254 CHAPTER 7: Similitude and Dimensional Analysis
7.28 (a) Use dimensional analysis and the MLT
system to arrange the following into a
dimensionless number: g, p, µ,, and a-.
( b) Name the dimensionless number.
7.29 (a) Use dimensional analysis and the MLT
system to arrange the following into a
dimensionless number: L , p, µ,, and a-.
( b) Name the dimensionless number.
7.30 Use dimensional analysis to derive an
expression for the height of capillary rise in
a glass tube.
7.31 Derive an expression for the drag on an
aircraft flying at supersonic speed.
7.32 Derive an expression for small flow
rates over a spillway, in the form of a
function including dimensionless
quantities. Use dimensional analysis with
the following parameters: height of
spillway P, head on the spillway H,
viscosity of liquid µ,, density of liquid p,
surface tension er, and acceleration due to
gravity g.
7.33 Derive an expression for the velocity of rise
of an air bubble in a stationary liquid.
Consider the effect of surface tension as
well as other variables.
8.1
CHAPTER 8
Steady Incompressible Flow
in Pressure Conduits
I n this chapter we will discuss some aspects of steady flow in pressure conduits.
We shall limit our discussion to incompressible fluids, that is, to those that
have p = constant. This includes all liquids. When gases flow with very small
pressure changes we can consider them incompressible, for then p = constant.
In this chapter we will assume isothermal conditions so as to eliminate thermo-
dynamic effects, some of which we discuss in Chap. 13.
LAMINAR AND TURBULENT FLOW
If we measure the head loss in a given length of uniform pipe at different veloc-
ities, we will find that, as long as the velocity is low enough to secure laminar
flow (Sec. 4.2), the head loss, due to friction, is directly proportional to the ve-
locity, as shown in Fig. 8.1. But with increasing velocity, at some point B, where
visual observation of dye injected in a transparent tube would show that the flow
changes from laminar to turbulent (Sec. 4.2), there will be an abrupt increase in
the rate at which the head loss varies. If we plot the logarithms of these two vari-
ables on linear scales or, in other wo~ds, if we plot the values directly on log-log
paper, we will find that, after passing a certain-· transition region (BCA in
Fig. 8.;t), the lines will have slopes ranging from about 1.75 to 2.00.
Thus we see that for laminar flow the drop in energy due to friction varies
as V, while for turbulent flow the friction varies as vn, where r, ranges from
about 1.75 to 2. The lower value of 1.75 for turbulent flow occurs for pipes with
very smooth walls; as the wall roughness increases, the value of n increases up to
its maximum value of 2.
The points in Fig. 8.1 are plotted directly from measurements made by
Osborne Reynolds (Sec. 7.4), and show decided curves in the transition zone
where values of n are even greater than 2. If we gradually reduce the velocity
from a high value, the points will not return along line BC. Instead, the points
will lie along curve CA~ We call point B the higher critical point, and A the
lower critical point.
However, velocity is not the only factor that determines whether the flow
is laminar or turbulent. The criterion is Reynolds number, which we discussed in
Sec. 7.4. For a circular pipe we usually take the significant linear dimension Las
255
 8.2 Critical Reynolds Number 257
256 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
the least disturbance will transform it instantly into turbulent flow. On the other
Log-log plot hand, it is practically impossible for turbulent flow in a straight pipe to persist at
values of R much below 2000, because any turbulence that occurs is damped out
by viscous friction. This lower value is thus much more definite than the higher
one, and is the real dividing point between the two types of flow. So we define
this lower value as the true critical Reynolds number. However, it may vary
slightly. It will be higher in a converging pipe and lower in a diverging pipe than
in a straight pipe. Also, it will be less for flow in a curved pipe than in a straight
one, and even for a straight uniform pipe it may be as low as 1000, where there
is excessive roughness. However, for normal cases of flow in straight pipes of
uniform diameter and usual roughness, we can take the critical value as

Rcrit = 2000 (8.2)

For water at 59°F (l5°C), using Eqs. (8.1) and (8.2) we find:
Transi~iori Turbulent
when D = 1 in (25 mm), ~rit = 0.30 fps (0.091 mis)
:
V and when V = 3 fps (0.91 mis), Dcrit = 0.10 in (2.5 mm)
Figure 8.1 We seldom encounter velocities or pipe diameters as small as these with water
Log-l?g plot for fiow in a uniform pipe (n = 2.00, rough-wall pipe; n = 1.75, smooth- flowing in practical engineering, though they may occur in certain laboratory .
wall pipe). instruments. Therefore, for such fluids as water and air, practically all cases of
engineering importance are in the turbulent-flow region. But if tfie fluid is a vis-
the diameter D, and thus from Eq. (7.6) cous oil, laminar flow often occurs.

DV '\:¥"A."'~~:-as"'t::~~~,:,..-1 1 :-v..--".M1.t~•~~•·~u--....,~,c"'fr,:!r.s.rc~~,,~vr,<""'~~~..:..1o,L•.li'Q.-.)lff'.:,-.C"M'~~1.1 ¥ . i " . J . ~ ~ ; . . ' \ " l ' . Q l . ' I " ~ ~ : ! . : " - ' - " ~ ~.•· ~

Pipe flow: (8.1)

JJ
~ SAMPLE PROBLEM 8.1
In a refinery oil (s = 0.85, v = 1.8 x 10- m ls) flows
5 2

through a 100-mm-diameter pipe at 0.50 Lis. Is the flow laminar or turbulent?

where V;le can use any consisknt system of units, ·because R is a dimensionless
number.1 Solution
4Q 4(0.0005 m3/s)
Eq. (4.7): V = 0.0637 mis
nD 2
n(O.l m) 2
8.2 CRITICAL REYNOLDS NUMBER
DV 0.10 m(0.0637 mis)
The u~per criti~al Reynolds number, corresponding to point B of Fig. 8.1, is Eq. (8.1): R= = 354
really mdetermmate _and depends on the care taken to prevent any initial dis- JJ 1.8 x 10- 5 m 2ls
turbance from aftect~ng the ~ow. Its value is normally about 4900, but experi- Using Eq. (8.2), since R < Rcrit = 2000, the flow is laminar. ANS
menters have mamta~ned lammar flow in circular pipes up to values of Ras high
as 50,000. However, m such cases this type of flow is inherently unstable, and

1 Some~imes we find it convenient to use a "hybrid" set of units and compensate with a
correction factor. Thus, by substituting V = Q/A and V = m/pA from Eq. ( 4.6) into Eq.
(8.1), we get . 8.2.1 Oil with a kinematic viscosity of 0.00015 ft2/sec is flowing through a 3-in-diameter
R = 1.273Q/vD = 1.273m/µ,D,
wher~ Q and _m are the v~lum_e flow rate and the mass flow rate, respectively. The last
f~rm is especially convement m ~he case of gases; it shows that in a pipe of uniform
di~meter the Reynol?s number is constant along the pipe, even for a compressible
flmd, where the density and velocity vary, if there is no appreciable variation in
temperature to alter the viscosity of the gas.
EXERCISES
pipe. Below what velocity will the flow be laminar?
8.2.2 Oil with a kinematic viscosity of 0.185 St is flowing through a 150-mm-diameter
pipe. Below what velocity will the flow be laminar?
8.2.3 Oil with a kinematic viscosity of 0.0035 ft 2/sec flows through a 4-in-diameter pipe
with a velocity of 15 fps . Is the flow laminar or turbulent?
 8.4 Friction Head Loss in Conduits of Constant Cross Section 259
258 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits

8.3 HYDRAULIC RADIUS, HYDRAULIC DIAMETER

y2;! Et hr
HGL
For conduits having noncircular cross sections, we need to use some value other
than the diameter for the linear dimension in the Reynolds number. The char- P1h
acteristic dimension we use is the hydraulic radius, defined as

(8.3)

where A is the cross-sectional area of the flowing fluid, and P is the wetted Datum
perimeter, that portion of the perimeter of the cross section where the fluid con-
tacts the solid boundary, and therefore where friction resistance is exerted on Figure 8.2
the flowing fluid . For a circular pipe flowing full,
n?- r D summation of forces acting on any fluid element must be equal to zero (i.e.,
Full-pipe flow: Rh= - (8.4)
2nr 2 4 L F = ma = 0). Thus, in the direction of flow,
Thus Rh is not the radius of the pipe, and so the term "radius" is misleading. If a (8.5)
circular pipe is exactly half full, both the area and the wetted perimeter are half
the preceding values; so R1i is r/2, the same as if it were full. But if the depth of where we define r0, the average s/zear stress (average shear force per unit area)
flow in a circular pipe is 0.8 times the diameter, for example, A = 0.674D 2 and at the conduit wall, by
P = 2.2lD, then R 11 = 0.304D, or 0.608r. We discuss part-full pipe flow further in
Sec. 10.7. (8.6)
The hydraulic radius is a convenient means for expressing the shape as well
as the size of a conduit, since, for the same cross-sectional area, the value of Rh
will vary with the shape. in which r 0 is the local shear stress 2 acting over a small incremental portion dP
In evaluating the Reynolds number for flow in a noncircular conduit of the wetted perimeter.
(Sec. 8.6), customarily we substitute 4Rh for Din Eq. (8.1). , Noting that sin a = (z 2 - z1 )/L and dividing each term in Eq. (8.5) by -yA
Workers in some engineering fields define 4Rh to be the hydraulic diameter gives
Dh. Then, for a pipe, Dh = D, which is fine, but Dh = 4Rh, which seems strange.
(8.7)

EXERCISES From the left-hand sketch of Fig. 8.2, we can see that the head loss due to fric-
8.3.1 What is the hydraulic radius of a 12 in by 16 in rectangular air duct? tion at the wetted perimeter is
8.3.2 What is the percentage difference between the hydraulic radii of a
300-mm-diameter duct and a 300-mm-square duct? h =
f
(z + P1) _(z
l '}' 2
+ P2)
'}'
'

8.4 FRICTION HEAD Loss IN CONDUITS
OF CONSTANT CROSS SECTION
This discussion applies to either laminar or turbulent flow and to any shape of
cross section.
Consider steady flow in a conduit of uniform cross section A, not necessar-
ily circular (Fig. 8.2). The pressures at sections 1 and 2 are Pi and p 2, respectively.
The distance between the sections is L. For equilibrium in · steady flow, the
This equation indicates that hf depends only on the values of z and p on the cen-
terline, and so it is the same regardless of the size of the cross-sectional area A.
Substituting hf for the right-hand side of this expression and replacing A/P in
2 The local shear stress varies from point to point around the perimeter of all conduits
(regardless of whether the wall is smooth or rough), except for the case of a circular
pipe flowing full where the shear stress at the wall is the same at all points on the
perimeter.
260 CHAPTER 8: Steady l11conzpressible Flow in Pressure Conduits 8.5 Friction in Cjrcular Conduits 261
Eq. (8.7) by Rh from Eq. (8.3), we get equation, we_may easily obtain an expression for the slope of the energy line,
- L
I1-J = To - (8.8) s = -
h1 ct v2
= --
Rhy
(8.12)
L Rh 2g
This equation is applicable to any shape of uniform cross section, regardless of ·
which we also know as the energy gradient.·
whether the flow is laminar or turbulent. Its derivation is very similar to that in
Later, in Sec. 8.12, we shall see that Eqs. (8.11) and (8.12) also apply to
Sec. 5.3, in which the conduit cross section was not constant, however.
rough-walled conduits.
F~r a smooth-walle~ conduit, where we can neglect wall roughness ( dis-
cussed m Sec. 8.10) , we might assume that the average fluid shear stress r at the
wall is some function of p, J..i , and V and some characteristic linear dim~nsion 8.5 FRICTION IN CIRCULAR CONDUITS
which we will here take as the hydraulic radius R1z. Thus '
In Sec. 8.3 we saw that for a circular pipe flowing full Rh = D/4. Substituting this
To = f(p , µ , V , Rh) (8.9) value into Eqs. (8.11) and (8.12), we obtain_(for both smooth-walled and rough-
Using the pi theorem of dimensional analysis (Sec. 7.7) to better determine the walled conduits) the well-known equ.a tion for pipe-friction head loss,
form of this relationship, we choose p, Rh, and Vas primary variables, so that
Circular pipe,
fI 1 = µpa 1R~1v c1 · flowing full L V2
(laminar or
h
f
=J--
D 2g
(8.13)
Il2 = 'fopa2Rt2v c2
turbulent flow):
With the dimensions of the variables being ML - 1 y - 1 for µ, ML - 1 y - 2 for T
3
ML - for p, L for Rh , and LT- 1 for V, the dimensions for IT 1 are 0 ht f v
s=- -
2
' .a nd = (8.14)
L D 2g
For M: 0 = 1 + a1
For L: where f = 4C1 = 8</>(R) (8.15)
0 = -1 - 3a 1 + b1 + c 1
For T: 0 = -1 - c 1 Equation (8.13) is known as the pipe-friction equation, and as the Darcy-
Weisbach equation. 3 Like the coefficient c1, the friction factor f ( also some-
The solution of these three simultaneous equations is a1 = b 1 = c = -1 from times called the Darcy friction factor) is dimensionless and some function of
1
which . '
Reynolds number. Much research has gone into determining how /varies with R
and also with pipe roughness (see Sec. 8.12). The pipe-friction equation states
rr 1-- _µ_ -- R - 1
that the head lost in friction in a given pipe is proportional to the velocity head.
pRhV
The equation is dimensionally homogeneous, and we may use it with any con-
_':h~re Rh Vp/µ is a Reynolds number with Rh as the characteristic length. In a sistent system of units. .
similar manner, we obtain Dimensional analysis gives us the prope,r form for an equation, but does
To not yield a numerical result, since it does not deal with abstract numerical fac-
Il2 = - tors. It also shows that Eq. (8.13) is a rational expression for pipe friction. But
pV2
we must determine the exact form of <f>(R) and numerical values for C1 and f by
Accor~ing to Sec. 7.7, step 7, we can write Il 2 = <f>(Il 11 ), which results in r = experiment or other means. ·
2
pV ¢(R). Setting the dimensionless term <f>(R) = ½C1 , this yields
0 For a circular pipe flowing full, by substituting Rh = r 0 /2, where r 0 is the ra- .
dius of the pipe, we can write Eq. (8.8) as ·
v2
CfpT (8.10) ht = To_f_ = 2roL (8.16)
Rh y 'oY
Inserting this value of r0 into Eq. (8.8), and noting from Eq. (2.l)·that y = pg, where the local shear stress at the wall, r 0, is equal to the average shear stress T0
because of symmetry.
L V2
h
f -
-C --
f Rh 2g (8.11)
3
In a slightly different form where the hydraulic radius Rh replaces D, so that/ changes,
which we can apply to any shape of smooth-walled cross section. From this Eq. (8.13) is known as the Fanning equation, which chemical engineers use widely.
 8.6 Friction in Noncircular Conduits . 263
262 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits

0 0.1 0.2 0.3 0.4 0 ;5 0 6 0 7 0 8 0 i 9 1 ;0

I I i i i FRICTION IN NONCIRCULAR CONDUITS
I I

- I I 8.6
-1
- - - I
I
I
Most closed conduits we use in engineering practice are of circular cross section;.
--- '" I
...... however, we do occasionally use rectangular ducts and cross sections of other
Yc _,,..
.'
~.,,I geometry. We can modify many of the equations in this chapter for application
- - I
I
to noncircular sections by using the concept of hydraulic radius.
In Sec. 8.3 we defined the hydraulic radius as Rh = A/P, where A is the
- - I ,.,.,,, I
I
cross-sectional area and Pis the wetted perimeter. Some equations include Rh,
I
- -v
1- -: - -i
Umax - - - - - - - - - - 1
as in Sec. 8.18, in which case we can simply determine R 11 for that conduit. But,
because circular pipes are so common, many more equations instead use the
Figure 8.3 pipe diameter, D. If the conduit of interest is not circular, we hav.e· no diameter
Velocity profile in laminar flow, and distribution of shear stress. to use in such equations, but we can find its hydraulic radius, R1i. Then we notice
for a circular pipe flowing full, that
· For a cylindrical fluid body of radius r concentric with the pipe, if we follow D (8.20)
a development similar to that of Eqs. (8.5)-(8 .8) and note that A = n? and p = 4
2nr, we can show that
(8.21)
h = 2TL (8.17) or
f ry

where r is the shear stress in the fluid at radius r; this also follows from This provides us with an equivalent diameter, which we can substitute into
Eq. (5.12). Since h is the same for all r (Sec. 8.4), equating the right sides of Eq. (8.13) for ex mple to yield
1
Eqs. (8.16) and (8.17) yields L V2 (8.22)
r = ro ,
r
(8.18) ht= f 4R~ 2g
0
and into Eq. (8.1) for Reynolds number to give
or t?e shear str~ss is zero at the center of the pipe and increases linearly with the
radms to a maximum value r 0 at the wall as in Fig. 8.3. This is true regardless of (4R i)Vp (4Rh)V (8.23) .
Pipe flow:
R = -1- - = - - -
whether the flow is laminar or turbulent. }.L V
From Eqs. (8.8) and (8.13) and substituting Rh= D/4 for a circular pipe we
obtain · · ' and into any other pipe flow equations and charts that use D instead of Rh. We
will meet many such equations and an important such chart later in this chapter,
t v
ro = - p -
2
t v2
-y- (8.19) and there are more in Chap. 13.
4 2 4 2g This approach gives reasonably accurate results for turbulent flow, but the
results are poor for laminar flow, because in such flow viscous action causes fric-
tional phenomena throughout the body of the fluid, while in turbulent flow the
With this equation, we can compute r 0 for flow in a circular pipe for any experi-
frictional effect occurs largely in the region close to the wall; i.e., it depends on
mentally determined value off.
the wetted perimeter.

EXERCISES
EXERCISES
8.5.1 Steam with _a specifi~ wei¥ht ?f 0.32 lb/ft3 is flowing with a velocity of 94 fps
through a circular pipe with J = 0.0171. What is the shear stress at the pipe wall? 8.6.1 When fluid of specific weight 46 lb/ft 3 flows in a 8-in-diameter pipe, the frictional
2
stress between the fluid and the pipe wall is 0.65 lb/ft . Calculate the friction head
8.5.2 Steam with a specific weight of 38 N/m3 is flowing with a velocity of 35 m/s loss per foot of pipe. If the flow rate is 1.8 cfs, hew much power is lost per foot of
through a circular pipe with f = 0.0154. What is the shear stress at the pipe wall?
pipe?
8.5.3 Oil (s = 0,?3) of viscosity 0.004 ft 2/sec flows in a 4-in-diameter pipe ~ta rate of 8.6.2 When fluid of specific weight 8.2 kN/m3 flows in a 150-mm-diameter pipe, the
2
6.5 gpm. Fmd the head loss per unit length. frictional stress between the fluid and the pipe wall is 25 N/m . Calculate the,
8.5.4 Oil (s = 0.92~ of viscosity 0.00038 m2/s flows in a 100-mm-diameter pipe at a rate friction head loss per meter of pipe. If the flow rate is 42 Lis, how much power rs
of 0.64 Lis. Fmd the head loss per unit length. lost per meter of pipe?
264 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits 8.8 Entrance Conditions in Laminar Flow 265
From this last equation, noting that y = gp and µ,/p = v, the loss of head due to
8.7 LAMINAR FLOW IN CIRCULAR PIPES
friction is
In Sec. 2.11 we noted that for laminar flow r = µ,du/dy, where u is the value of
the velocity at a distance y from the boundary. As y = r0 - r, we also see that Laminar fL L L
flow:
h1 = 32--V = 32v-V (8.28)
r = -µ, du/dr; in other words, the minus sign indicates that u decreases as r in- 'Y D2 gD2
creases. The coefficient of viscosity µ, is a constant for any particular fluid at a
constant temperature, and therefore if the shear varies from zero at the center which is the Hagen-Poiseuille law for laminar flow in tubes. G. Hagen, a German
of the pipe to a maximum at the wall, it follows that the velocity profile must engineer, experimented with water flowing through small brass tubes, and pub-
have a zero slope at the center and have a continuously steeper velocity gradient lished his results in 1839. J. L. Poiseuille, a French scientist, experimented with
approaching the wall. water flowing through capillary tubes in order to determine the laws of flow of
To determine the velocity profile for laminar flow in a circular pipe, we blood through the veins of the body, and published his studies in 1840.
substitute the expression r = µ,du/dy into Eq. (8.17). Thus From Eq. (8.28) we see that in laminar flow the loss of head is proportional
2rL du 2L du 2L to the first power of the velocity. This is verified by experiment, as shown in
hf = ry = fL dy ry = - fL dr ry Fig. 8.1. The striking feature of this equation is that it involves no empirical co-
efficients or experimental factors of any kind, except for the physical properties
h(Y of the fluid such as viscosity and density (or specific weight). From this, it would
From this, du ---rdr appear that in laminar flow the friction is independent of the roughness of the
2µ,L
pipe wall. Experiments also bear out that this is true.
Integrating and determining the constant of integration from the fact that Dimensional analysis shows that we may express the friction loss by
u = umax when r = 0, we obtain Eq. (8.13). Equating (8.13) and (8.28) and solving for the friction factor f, we
obtain for laminar flow under pressure in a circular pipe,
h1Y 2
U = llmax - 4µ,Lr = llmax - kr2 (8.24)
Laminar
flow: (8.29)
From this equation we see that the velocity profile is a parabola, as shown in
Fig. 8.3. Note that k = h1y/4µ,L. . ..
At the wall we have the no-shp boundary condition (Sec. 2.11) that u = 0 Thus, if R is less than 2000, we may use Eq. (8.28) tp find pip~ friction)lead loss,
when r = r0 . £ubstituting this into the second expression of Eq. (8.24) and no.t- or we may use the pipe-friction equation (8.13). with the value off as given by
ing tha~ umax = ~, the centerline velocity, we find k = ~/r5. Thus we can express Eq. (8.29).
Eq. (8.24) as
EXERCISES
(8.25)
8.7.1 An oil with kinematic viscosity of 0.004 ft 2/sec weighs 62 lb/ft 3. What will be its
flow rate and head loss in a 2750-ft length of a 3-in-diameter pipe when the
Reynolds number is 950?
Combining Eqs. (8.24) and (8.25), we get an expression for the centerline veloc-
ity as follows 8.7.2 With laminar flow in a circular pipe, at what distance from the centerline (in
terms of the pipe radius) does the average velocity occur?
h1Y h1Y 2
1,
Ve = Umax =
--?=--D
4µ,L O 16µ,L
(8.26) 8.7.3 For laminar flow in a two-dimensional passage, find the relation between the
average and maximum velocities.
We can multiplyEq. (8.24) by a differential area dA = 2nrdr and integrate
the product from r = Oto r = r0 to find the rate of dis~harge. From Eq. (4.3), t~e 8.8 ENTRANCE CONDITIONS IN LAMINAR FLOW
rate of discharge is equivalent to the volume of a solid bounded by the velocity
profile. In this case the solid is a paraboloid with a ~aximu~ height of umax· The In the case of a pipe leading from a reservoir, if the entrance is rounded so as
mean height of a paraboloid is one-half the maximum height, and hence the to avoid any initial disturbance of the entering stream, all particles will start to
mean velocity Vis O.Sumax· Thus flow with the same velocity, except for a very thin film in contact with the wall.
Particles in contact with the wall have zero velocity (the no-slip boundary
V = -h1Y
- D2 (
8.27 ) condition, Sec. 2.11), but the velocity gradient there is extremely steep, and, with
32µ,L this slight exception, the velocity is uniform across the diameter, as shown in
266 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
8.8 Entrance Conditions in Laminar Flow 267
~deal, originated in 1904 with Ludwig Prandtl (1875-1953) G ·
,.,,-- Rounded entrance · ,,,.- Inviscid core Fully developed mg profe p h h . 'a erman engmeer-
I from tank /
A parabolic profile . _th· ssor. e! aps ~ e smgle most significant contribution to fluid mecha
1cs,th is _cosncept is particularly important with turbulent flow; we will discuss~;
f ur er m ecs. 8.10-8.12 and Chap. 9.
As ~e saw in Sample Prob. 5.1 for a circular pipe the kinetic ene a
stream with a parabolic velocity profile is 2V3/2g (becau~e rv - 2) h . rvb~ ofl a
Boundary layer B mean vel ·t At th . ~ - , w e1e Is t 1e
-- Entrance length, Le _ _ ____ ___ .., dia t oci y. f e entrance to the pipe the velocity is uniformly V across the
me er, excel_)t or an extremely thin layer next to the wall Thus at th
-------.-- - - - Unestablished flow, entrance flow - - -- ---Established flow trance to the th k · · · e en-
the ct· t z1~e ~. metic_ energy_ per unit weight is practically V3/2g. So in
IS ance e t ere IS a contmuous mcrease in kinetic ener accom ·
Figure 8.4
a correspo~ding d~crease in press_ure head. Therefore, at a ctf!tance L p~:I:dtt~
Velocity profiles and development of the boundary layer along a pipe in laminar flow.
:::rr:~~~v::~~~f ~~/~g
~~:; r:i:~70:~~;~ i~e~~:~::::0t~~n the stati~ value in ~~:
Fig. 8.4. As the fluid progresses along the pipe, friction originating from the wall W~ have dealt _with laminar flow rather fully, not merel because it i .
ft;:~:i:~ ~r~i~~~s ~n~olving fluids of very high viscosity, but ~specially bee:~~
0
slows down the streamlines in the vicinity of the wall, but since Q is constant for
successive sections, the velocity in the center must accelerate, until the final veloc-
ity profile is a parabola, as shown in Fig. 8.3. Theoretically, this requires an infinite also help with the stu~y
rigid mathematical treatment is impossible.
il~~:::~e:~tA~~lw~:;~s~~n!~~swns i
general approach will
are so complex that
distance, but both theory and observation have established that the maximum
velocity in the center of the pipe will reach 99% of its ultimate value in a distance4
• ~.t:\ ~~s:x.::.-;..'\iJ.L,...,,,. , _ _
,. "'.!_..?J:·.,....~~-o\r,! :.;..c.: ....J:;,c:~!f""':~.: .....tC\:r~r;.;
Le = 0.058RD (8.30) , •... "1"' ,,...it~• ~ c.:;. ,...,:,~..,,_. ~ -:ic«rc

' SAMPLE PROBLEM 8.2 For the case of S l p '"· -._

We call this distance the entrance length (Fig. 8.4). For the critical (maximum)
velocity, the velocity at r = 20 mm th f . t' a f e rob. 8.1, find the centerline
value of R = 2000 (Sec. 8.2), the entrance length Le equals 116 pipe diameters. wall, and the head loss per meter ~f P~P~Ife::ih~ctor, the shear stress at the pipe
In other cases of laminar flow with Reynolds numbers less than 2000, the dis-
tance Le will be correspondingly less, in accordance with Eq. (8.30). Solution
Within the entrance length the flow is unestablished; that is,_the velocity From the solution to Sample Prob. 8.1: The flow is laminar.
profile is changing. In this region (Fig. 8.4), we can visualize the flow as consist- Th~refore, per Sec. 8.7: l{ = 2V = 0.1273 mis ANS
ing of a central inviscid core in which there are no frictional effects, i.e., the flow
is uniform, and an outer, annular zone extending from the core to the pipe wall. Eq. (8.24): = kr 2
U Umax -- U max
· = T7
Ve = 0•1273 mis
This outer zone increases in thickness as it moves along the wall, and is known
as the boundary layer. Viscosity in the boundary layer acts to transmit the effect When r = ro = 50 mm, u = O; hence O = 0.1273 - k(0.05)2, from which
of boundary shear inwardly into the flow. At section AB the boundary layer has k = S0.9/(m·s) Uzomm = 0.1273 - 50.9(0.02) 2 = 0.1070 mis
grown until it occupies the entire cross section of the pipe. At this point, for lam- ANS
inar flow, the velocity profile is a perfect parabola. Beyond section AB, for the Eq. (8.29): 64 64
same straight pipe the velocity profile does not change, and the flow is known as t = R = 354 = o.1810 ANS
(laminar) establislzedflow, or (laminar)/u/ly developedflow. The flow will con-
tinue as fully developed so long as no change occurs to the straight pipe surface. ~ Eq. (8.19): ' 0 = f
-p-
v2 = 0.1810 co 0637 1 ) 2
--(850k I 3) · ms
When a change occurs, such as at a bend or other pipe fitting, the velocity pro- %
):
4 2 4 gm 2
file will deform and will require some more flow length to return to established ~ = 0.0779 kgl(m·s 2)
flow. Usually such fittings are so far apart that fully developed flow is common; kg N· 2
but when they are close enough it is possible that established flow never occurs. 0.0779---s_ = 0.0779Nlm 2 ANS
The concept of a boundary layer within which viscosity is important, and (m·s 2) kg·m
;: .
outside of which friction is unimportant and we can consider the fluid to be h1 l V2
Eq. (8.14): s _ _1_ (0.0637 mls) 2

4 H.L. Langhaar, Steady Flow in the Transition Length of a Straight Tube, J Appl.
L = f D 2g
0.000374 mlm
= 0 1810
0.10 m 2(9.81 2) mls
ANS
Mech., Vol. 10, p. 55, 1942. ·
268 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits 8.9 Turbulent Flow 269
merge into adjacent eddies. Thus there is a continuous mixing of particles, with
EXERCISES a consequent transfer of momentum. Viscosity dissipates mechanical energy,
8.8.1 In Exer. 8.2.3 what will be the approximate distance from the pipe entrance to the generating small amounts of heat.
first point at which the flow is established?
8.8.2 In Exer. 8.5.4 what will be the approximate distance from the pipe entrance to the -First Expression
first point at which the flow is established?
In the modern conception of turbulent flow, we assume a mechanism similar to
that just described for laminar flow. However, we replace the molecules by
minute but finite masses or eddies (Fig. 8.5c ). So, by analogy, for turbulent flow
8.9 TURBULENT FLOW
we may define the shear stress along the plane through ab in Fig. 8.5 as
In Sec. 4.2 we saw that in lamina1:. flow the fluid particles move in straight l~nes
while in turbulent flow they follow random paths. Consider the case of lammar du
Turbulent shear stress = 17- (8.31)
flow shown in Figs. 8.. 5(a, b), where the velocity u_increases)Yith y. Even though dy
the fluid particles are moving horizontally to the nght, becau~e of molecular mo-
tion, molecules will cross line ab and transport momen~um w~th them. On _the a:- But unlikeµ.,, the eddy viscosity 1J (eta) is not a constant for a given fluid at a given
erage, the velocities of the molecule_s in th~ slower-movmg flu~d below the lme will temperature, but it depends on the turbulence of the flow. We may view it as a co-
be less than those of the faster-movmg flmd above; the result 1s t~at th_e m~lecules efficient of momentum transfer, expressing the transfer of momentum from points
that cross from below tend to slow down the faster-moving flmd. L1kew1se, ~he where the velocity is low to points where it is higher, and vice versa. Its magnitude
molecules that cross the line ab from above tend to speed up the slower-m~vmg may range from zero to many thousand times the value ofµ.,. However, its numer-
fluid below. The result is a shear stress r (tau) along the surface wh~se trace 1s ab. ical value is of less interest than its physical concept. In dealing with turbulent flow
As given in Sec. 2.11, r = µ,du/dy. This equat~o1:1- is applicable to lammar flow onlr it is sometimes convenient to use kinematic eddy viscosity e ( epsilon) = 17/p,
Let us examine some of the charactenstics of turb~~ent flo":' to_ see how 1t which is a property of the flow alone, analogous to kinematic viscosity.
differs .from- laminar flow. In turbulent flow the velocity at a-p01!:t m the flow In general, the total shear stress in turbulent flow is the sum of the laminar
field fluctuates in 60th magnitude and direction. 5 We may observe th~se fluctu- shear stress plus the turbulent shear stress, i.e.,
ations in accurate velocity measurements (Sec. 11.4), and ":'e commonly see
du du du
their effects on pressure gages and manometers. Th~ fluctuations result fron~ a r = µ.,- + 17- = p(v + e)- (8.32)
multitude of small eddies (Sec. 4.2), created by the viscous shear bet~een ~dJa- dy dy dy
cent particles. These eddies grow in size and .then disappear as their particles With turbulent flow the second term of this equation is usually many times
larger than the first term.
5 We can best visualize the velocity at a point in a so-called "steady" turbulent flo~ as a For turbulent flow we saw in Sec. 4.5 and Fig. 4.6 that the local axial veloc-
vector that fluctuates in both direction and magnitude. The mean temporal velocity at ity has fluctuations of plus and minus u', and there are also fluctuations of plus
that point corresponds to the "average" of those vectors. and minus v' and w' normal to u as shown in Fig. 8.6b. As it is obvious that there

y
u + du

lib
! +v'
u + Liu Jy ~ cj jdy
a--rrTrr- d
u'
T ll Mean velocity u
~ b b +u'
L1y a- du
J_ u
- -0- - - -0-- - - -O- -
a~
Velocity profile
I I
lu'I ( +w'', ,'
--O-D--0 -v'
du li71
dy e Velocity at a particular instant =

Velocity u - o Molecule Q Finite fluid V(u :±: u') 2 + (v') 2 + (w')2

mass or eddy f = 1i7(,du
dy
(b) (c)
(a) (a) (b)

Figure 8.5 Figure 8.6

(a) Velocity profile. (b) Laminar flow (transfer of molecules across ab). (c) Turbulent
(a) Mixing length t. (b) Instantaneous local velocity in turbulent flow.
flow (transfer of finite fluid masses across ab).
270 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
can b.e no value~ of v' next to and perpendicular to a smooth wall, turbulent flow
cannot exist there. Hence, near a smooth wall, the shear is due to laminar flow
alone; and r = µ,du/dy. Note that the shear stress always acts to cause the veloc-
ity distribution to become more uniform.
At some distance from the wall, such as 0.2r, the value of du/dy becomes
small in turbulent flow, and so the viscous shear becomes negligible in compari-
son with the turbulent shear. The latter can be large, even though du/dy is small,
because of the possibility of T/ being very large. This is because of the great tur-
bulence that may exist at an appreGiable distance from the wall. But at the cen-
ter of the pipe, where du/dy is zero, there can be no shear at all. So, in turbulent
flow as well as in laminar flow, the shear stress is a maximum at the wall and de-
creases linearly to zero at the axis, as shown in Fig. 8.3 and proved in Sec. 8.5.
Second Expression
We can obtain another expression for turbulent shear stress that is different
from Eq. (8.31). In Fig. 8.5a, if a mass m of fluid below ab, where the temporal
mean axial velocity is u, moves upward into a zone where the temporal mean
axial velocity is u + Liu, its initial momentum in the axial direction must increase
by mL1u. Conversely, when a mass m moves from the upper zone to the lower its
axial momentum will decrease by mL1u. So this transfer of momentum back and
forth across ab will produce a shear in the plane through ab proportional to L'.lu.
This shear is possible only because of the velocity profile shown. If the latter
· were vertical Liu would be zero and there could be no shear.
If the distance L1y in Fig. 8.Sa is chosen so that the average value of +u' in
the upper zone over a time period long enough to include many velocity fl~ctu-
ations is equal to Liu, i.e., Liu = lu' I, the distance between the two streams wi!l be
what is known as the mixing .length f, which we will discuss shortly. Consider,
over a short time interval, a mass moving upward from below ab. with a veloc-
ity v'; it will transport into the upper zone, where the velocity is u + u', a
momentum per unit time which is on the average equal to p(v'dA)(u). Th_e
slower-moving mass from below ab will tend to retard the flow above ab; this
creates a shear force along the plane of ab. We can find this force by applying the
momentum principle, Eq. ( 6.6),
F = rdA = pQ(L1V) = p(v'dA)(u + u' - u) = pu'v'dA
So, over a period of time sufficiently long to include a large number of velocity
fluctuations, the shear stress is given by
r = F/dA -pu'v' (8.33)
where u'v' is the temporal average of the product of u' and v'. This is an alter-
nate form for Eq. (8.31), and .in modern turbulence theory -pu'v' is known as
the Reynolds stress. _ file _changes because of the growth of a laminar boundary layer, which continues
The minus sign appears in Eq. (8.33) because the product u'v' on the aver-
age is negative. By inspecting Fig. 8.Sa, we can see that +v' is associated with
-u' values more than with +u' values. The opposite is true for -v'. Even
though the temporal mean values of u' and v' are individually equal to zero, the
8.10 Viscous Sublayer in Turbulent Flow 271
temporal mean value of their product is not zero. This is because combinations
of +v' and -u' and of -v' and +u' predominate over combinations of +v' and
+u' and -v' and -u', respectively. ~
Pr~ndtl (Sec. 8.8) reasoned that in any turbulent flow ~ and /v'i
must be
proport10nal to each other and of the same order of magnitude. He also intro-
duced the concept of mixing length f, which is the distance transverse to the flow
direction such that L'.lu =~-From Fig. 8.6a we can see that L'.lu = f du/dy and
so /u'I = f du/dy. If /u'/ ex: /v'/ and if we allow f to account for the constant of
proportionality, Prandtl 6 showed that - u' v' varies as f 2 ( du/dy ) 2 . Thus
2
r = du)
-pu'v' = pf 2 ( dy (8.34)
This eq~ation ex~ress:s ~erms that we can measure. Thus in any experiment that
r
determmes the pipe ~nct10n loss , we can compute 0 from Eq. (8.8), and then we
c~n find r a~ an_Y radms from E_q. (8.18). A traverse of the_velocity across a pipe
dia~eter will give u at any radms, and the velocity profile will give du/dy at any
r~dms.1:'hus Eq. (8.34) allows us to find the mixing length eas a function of the
pipe radms. ~he purpo~e of all of this is to help us develop theoretical equations
for_ the veloc_ity profile m turbulent flow, and from this in turn to develop theo-
retical equat10ns for f, the friction coefficient.
EXERCISES
8.9.1 Tests on 70°F water flowing through a 9-in-diameter pipe showed that when
V = 13 fps, f = 0.0162. (a) I!, at a distance of 3 in from the center of the pipe,
r = 0.38_8 psf, and t~e velocity profile gives a value for du/dy of 6.971sec, find at
that radms (a) the VISCOUS shear, (b) the turbulent shear, and ( C) the mixing
length f.
8.9.2 Wate~ at 20°C flows through a 240-mm-diameter pipe. Tests have determined that
at a distance o~ 60 mm from the pipe centerline the mixing length f·is 19.8 mm and
from the velocity pr~file ~u/dy = 5.33 s- 1. Find at that radius (a) the total shear
stress, (b) the eddy v1scos1ty, (c) the viscous shear, and (d) the turbulent shear.
8.10 Viscous SUBLAYER IN TURBULENT FLOW
I~ Fig. 8.4 we saw that, for laniinar flow, if the fluid enters a pipe with no initial
?isturb~nce, the velocity is uniform across the diameter except for an exceed-
m_gly thu~ ~lm at the wall, because the velocity touching any wall is zero (the no-
slip condit10n, Sec. 2.11). But as flow proceeds down the pipe, the velocity pro-
until ~he boundary_ layers from opposite sides meet at the pipe axis and then
there 1s fully developed ( or established) laminar flow.
6
H. Schlichting, Boundary Layer Theory, 7th ed., McGraw-Hill Book Co., 1987, p. 605.
272 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
/ Laminar boundary layer
y
'
r
Fully developed
turbulence
Figure 8.7
D evelo pment of boundary layer in a pipe where full y d eveloped fl ow is turbulent
(scales much di storted).
If the Reynolds number is above the critical value (Sec. 8.2) , so that the de-
veloped flow is turbulent, the initial condition is much like that in Fig. 8.4. But as
the laminar boundary layer increases in thickness, at a certain point a transition
occurs and the boundary layer becomes turbulent (Fig. 8.7).·This transition oc-
curs where the length xc of the laminar portion of the boundary layer is about
equal to 50b 1000v/V, where Vis the uniform velocity (i.e. , where Rx = Ux/v =
500,000). After the transition , the t(!,rbulelit boundary layei· generally increases
in thickness much more rapidly. So the length of the inviscid core, to where the
two opposite layers meet , is relatively shorter.
The development of turbulent flow is considerably more complex than that
of laminar flow (Sec. 8.8). Although the length of the inviscid core is relatively
short with turbulent flow, it takes about four times this length for the velocity
profile lo become fully developed, and 8 to 12 times this length for the detailed
structure of the turbulence to become fully developed. Only when all these as-
pects are complete do we have fully developed turbulent flow. These various de-
velopments depend on a variety of features, including the entrance conditions
(Sec. 8.21) and the wall roughness (Secs. 8.10-8.12); as a result no single equation
exists to predict the entrance length for turbulent flow, i.e., the length over which
the flow is developing. But as an approximate guide, we may observe that the ve-
locity profile is generally fully developed within 20 to 40 pipe diameters.
In all that now follows we shall consider only fully developed turbulent
flow.
As v' must be zero at a smooth wall, turbulence there is inhibited so that a
laminar-like sublayer occurs immediately next to the wall. However, the adja-
cent turbulent flow does repeatedly induce random transi~nt effects that mo-
mentarily disrupt this sublayer, even though they fade rapidly. Because it is
therefore not a true laminar layer, and. because shear in this layer is predomi-
nantly due to viscosity alone , we call it a viscous sublayer (see Fig. 8.7). This vis-
cous sublayer is extremely thin , usually only a few hundredths of a millimeter,
but its effect is great because of the very steep velocity gradient within it and
because r = µ.,du/dy in that region. At a greater distance from the wall the
8.10 Viscous Sublayer in Turbulent Flow 273
y
zone
,..__----...u
Figure 8.8
Velo~ity pro~le ~ear a·solid wall (vertical scale greatly exaggerated). Theoretical
relat10ns (sohd Imes) are compared with experimental data (small circles).
viscous effect becomes negligible, but the turbulent shear is then large. Between
the two,_ ther_e must be a transition zone where both types of shear are signifi-
cant. It 1s evident _that there cannot be sharp lines of demarcation separating
these three zon~s, mstead one must merge gradually into the other.
_By p~otting o~e velocity profile from the wall on the assumption that the
flow is_ entirely lammar (Sec. 8.7) and plotting another velocity profile on the as-
sumption that the flow is entirely turbulent (Sec. 8.11), the two will intersect as
sh?wn _in Fi~. 8.8. It i_s obvious that there can be no abrupt change in profil~ at
t~1s pomt of !ntersect10n, but that one curve must merge gradually into the other
w:1t~ some km? of transition, which is in fact what happens as shown by the ex-
penmental pomts.
When studying sue~ velocity profiles, we find that th_e quantity ~ fre-
9uently occurs. Because 1t has the dimensions of velocity, scientists have named
it the shea_r-stress velocity (or friction velocity) u*' although it is not actually a
flow velocity. · ·
~hen the flow in a circular pipe is entirely laminar, we have seen that the
velocity profile is a parabola (Fig. 8.3). But when there is only an extremely thin
fil1!1 closest to ~he wall w~e~e viscous shear dominates, we can scarcely distin-
gmsh t~e ve~oc1t~ pr_ofile m 1t from a straight line. If we ignore the momentary
fluctuations m this viscous sublayer, with such a linear velocity profile Eq. (2.9)
at the wall becomes
u To vu
ro = µ- or
y p y
 8: Steady Incompressible Flow in Pressure Conduits
8.10 Viscous Sublayer in Turbulent Flow 275
274 CHAPTER

But from the definition of u*, we have u; = 1:0/p, so, by eliminating r 0/p, we
obtain
vu
= u*2 (a)
y
u yu* (8.35)
or 1)
u*

which is known as a law of the wall. This linear relation for u(~) _appro~imates (b)
experimental data well in the range O ~ yu*/v ~ 5. If_we call this;m~recise, b~t Figure 8.9
commonly accepted, upper limiting value of y the thickness oft e viscous su - Turbulent flow near a boundary. (a) Low R, 8v > e; the pipe behaves as a smooth pipe.
(b) Relatively high R, 8v < e; if 8v < -Ae, the pipe behaves as a fully rough pipe.
layer, 81,, then
(8.36)

The transition zone appears to extend from a to c in Fig. 8.~. For the l~tter surface roughness, the roughness has no effect on friction, and the pipe is hy-
point, the value of y seems t~ be ab_o~t ?Ov/u* or 148v. Beyond this, the flow is so draulically smooth. If eu,)v > 70 ( or e > 148J, the pipe will behave as fully
turbulent that viscous shear 1s negligible. rough. In the region between these values, i.e., when the roughness projections
Noting from Eq. (8.19) that are such that 5 ~ eu*/v ~ 70 (or 8v ~ e ~ l48v), the pipe will behave in a transi-
tional mode, neither hydraulically smooth nor fully rough. Most engineering
pipe flows fall in this range.
u = {¼=V[l_ (8.37)
* \J-;; \Js Because the thickness of the viscous sublayer in a given pipe decreases
with an increase in Reynolds number, we see that the same pipe may be hy-
draulically smooth at low Reynolds numbers and rough at high Reynolds num-
and that the Reynolds number R = DV/v, we see that when yu*/v = 5, or Y = 0v, bers. Thus even a relatively smooth pipe may behave as a rough pipe if the
Reynolds number is high enough. It is also apparent that, with increasing
14.14v 14.14D (8.38) Reynolds number, there is a gradual transition from smooth to rough-pipe flow.
8v = Vv} RVf Figure 8.9 depicts these concepts schematically. '

From this, we see that the higher the velocity or t?e lower the ki~ema~ic viscosity,
the thinner is the viscous sublayer. Thus, for a given constant pipe drn~eter, the
thickness of the viscous sublayer decreases as the Reynolds number mc~easel~-
N ow we can consider what is meant by a smooth wall and a roug _wa ·
There is no such thing in reality as a mathematically smooth surface. But if the
irregularities on any actual surface are small enough that_ the effects of t~e pr~-
.ections do not pierce through the viscous ~ubl~yer (~ig. 8.8), the sur ace is
j1
ydraulically smooth from the fluid-mechamcs _v1ewpomt: If the effects o~ t~e
ro ·ections extend beyond the sublayer, the lammar layer is broken up ~n . t e
;uriace is no longer hydraulically smooth. If the sur!~ce rough~e~s pro1ectwns
are laroe enough to protrude right through the trans1t1on layer, 1t 1s totally ~ro-
0
ken up The resulting flow is completely turbulent, known asfully ~ough (-piP_e)
flow a~d friction is independent of Reynolds number. We shall discuss the_ s1g-
nifi;ance of this in Sec. 8.11. If the roughness projections protrude only parti~lly
into the transition layer, we say the flow is transitionally rough, and there 1s a
moderate Reynolds number effect. . . _
To be more specific, if e is the equivalent height of the roughness ~roJeC
· h f / < 5 (or e < 8 ) the viscous sublayer completely bunes the
t1ons t en or eu* v v
EXERCISES
8.10.1 Compute 8v for the data of Sample Prob. 8.3.
8.10.2 Water in a pipe (f = 0.018) is at a temperature of 70°F. (a) If the mean velocity
is 14 fps, what is the nominal thickness 8i, of the viscous sublayer? (b) What will
8v be if we increase the velocity to 24 fps and f does not change?
8.10.3 Water in a pipe(/= 0.012) is at a temperature of 15°C. (a) If the mean velocity
is 3.2 mis, what is the nominal thickness ov of the viscous sublayer? (b) What will
8v be if we increase the velocity to 5.5 mis and f does riot change?
8.10.4 For the data of Exer. 8.10.2(a), what is the distance from the wall to the assumed
limit of the transition region where true turbulent flow begins?
8.10.5 For the data in Exer. 8.10.3(a), what is the distance from the wall to the assumed
limit of the transition region where true turbulent flow begins?
8.10.6 Water at 50°C flows in a 150-mm-diameter pipe with V = 6.5 mis and
e = 0.14 mm. Head loss measurements indicate thatf = 0.020. (a) What is the
thickness of the viscous sublayer? (b) Is the pipe behaving as a fully rough pipe?
276 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
8.11 Velocity Profile in Turbulent Flow 277
8.11 VELOCITY PROFILE IN TURBULENT FLOW However, this discrepancy is well wjthin the c .
Prandtl (Secs. 8.8 and 8.9) reasoned that turbulent flow in a pipe is strongly in- wh~re the equation is not supposed to a 1 , o;1111es of the viscous sublayer,
fluenced by the flow phenomena near the wall. In the vicinity of the wall, r = the mtervening transition or overlap zon~p/ ' a~ where we have_ Eq. (8.35). In
bulent shear are important invest" rt h( ig. .8), where both viscous and tur-
r 0 . He assumed that the mixing length f (Sec. 8.9) near the wall was propor-
profile data follow a logariihmic r;f:ti~~ ave found that experimental velocity
tional to the distc\nce from the wall, that is, f = Ky. Experiments have con-
firmed this , and have determined that K = 0.40. Using this relationship in Eq.
(8.34), we get
-u = 2.5111 (yu)
-* + 5.0
r = r0 = pc (!;J
2
= pK
2
y2(!;)
2 U* V

Although Eq. (8.40) is not perfect it reliabl fi h

(8.41)

si:nan areas mentioned, where it is still close Sy ts t e data exc~pt for the two
or du= ~ft:=~; discharge Q with a high deg
E
f . owe can determme the rate of
ree o accuracy by ·
q. (8.40) and integrating over the area of th
h
. usmg t e value of u given by
e pipe. Thus
Integrating, and inserting 0.4 for K, we obtain

u = 2.Su*lny + C
Q = fudA = 2nfurdr
0

We can evaluate the constant C by noting that u = umax = the centerline veloc- ~ubstituting from the first expression of E (8 4 . .
mg by the pipe area n,2 the mean 1 ·t q. 7 . 0) for u, mtegratmg and divid-
ity when y = r 0 = the pipe radius. Substituting the expression for C, we get . 0, ve OCI y IS

Umax -
--- =
U ro
2.5ln- (8.39) V -_ umax - 2.Su* [ lnro - r6 lro
2 o rln(ro - r)dr ]
u* y
Makino0 use of Eq · (8 · 37) , th·is equat10n
. reduces to
This is known as the velocity defect law, because we call ( umax - u) the velocity
defect. Replacing y by r0 ~ r, and changing the base e logarithm (ln) to a base 10
V=u max _3
2X
25
= l •326V"'VT
lrf (8.42)
logarithm (log), the equation becomes · U*, U max -

From Eq. (8.42) we can obtain the i e fi • . .

(8.40) the maximum velocity. It is pp actor, which is the rat10 of the mean to

Although this equation is derived by assuming certain relations very near to the
wall, it holds almost as far out as the pipe axis.
Pipe factor = V 1
1 + 1.326\lj (8.43)
Starting with the derivation of Eq. (8.34), this entire development is open
to argument at nearly every step. But the fact remains that Eq. (8.40) agrees Using Eq. (8.43) to eliminate u fr E ( .
very closely with actual measurements of velocity profiles for both smooth and nate Lt *' the resu It 1s
· . max om q. 8 .40) and usmoo Eq · (8 •37) to e 1·1m1-
·
rough pipes. However, there are two zones in which the equation is defective. At
the axis of the pipe, du/dy must be zero. But Eq. (8.40) is logarithmic and does
not have a zero slope at r = 0, and hence the equation gives a velocity profile
with a sharp point ( or cusp) at the axis, whereas in reality it is rounded at the
u = (l + l.326Vf)V - 2.04VfVlog-ro_
ro - r (8.44)
axis. This discrepancy affects only a very small area and causes only a very slight
error when computing the rate of discharge using Eq. (8.40).
Equation (8.40) is also not applicable very close to the wall. In fact, it indi-
cates that when r = r 0 , the value of u is minus infinity. The equation indicates 7 The _integral results in indeterminate values at r - r
equat10n for u does not really a 1 1 - o, as we should expect. since the
that u = 0, not at the wall, but at a small distance from it, shown as y1 in Fig. 8.8. PP Yc ose to the wall How f .
purposes, these reduce to neglio1'ble t· . . ever, or a11 practical
e, quan 1ties.
278 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
8.11 Velocity Profile in Turbulent Flow 279
r~--
: S:\-M~LE PROBLEM 8.3 . The pipe friction head loss in 200 ft of 6-in-diameter
• pipe 1s 2_5 ft· lb/lb when 011 (s = 0.90) of viscosity 0.0008 lb·sec/ft 2 flows at 2.0 cfs.

-- ·- -·-·-·-·-· -, --·
I
I Determn:-e the c_enterline velocity, the shear stress at the wall of the pipe, and
the velocity at 2 m from the centerline.
t Solution
Rough pipe I First determine whether the flow is laminar or turbulent: ·
R= 107.f =0.04 e
I Eq. (4.7):
4
V = Q2 =
4 2
( 0) = 10 19 fps

I'
~ nD n(0.5) 2 •

Figure 8.10 DVp 0.5(10.19)(0.9 x 1.940)

. Eq. (8.1): R = -
Velocity profiles across a pipe for equal flow rates. The turbulent profiles are plotted 11,120
from Eq. (8.44).
I
I
Since R
µ,
> 2000, the flow is turbulent.
0.0008

which enables us to plot a velocity profile for any mean velocity and any value of
I From Eq. ( 8 .l3): f = hrD(2g) = 25(0.5)2(32.2)
0.0388
! L V2 200(10.19) 2
fin turbulent flow. In Fig. 8.10 profiles for both a smooth and a rough pipe are
plotted from this equation. The only noticeable difference between these and
measured profiles is that the latter are more rounded at the axis of the pipe. 8 Of
l
~

I
From Eq. (8.43), um,, = 10.19(1 + l.326\10.0388) = 12.85 fps ANS
course, the measured turbulent profiles also exhibit turbulent fluctuations
~ E
2
( )· = fp v _ o.0388(0.9 x 1.940)(10.19) 2
everywhere except near the walls. . q. 8 .19 . To 8 - 8 0.878 lb/ft 2 ANS
Comparing the turbulent-flow velocity profiles with the laminar-flow
velocity profile in Fig. 8.10 we see the turbulent-flow profiles are much flatter
near the central portion of the pipe and steeper near the wall. We also notice that
the turbulent profile for the smooth pipe is flatter near the center (i.e., blunter)
Ii Eq. (8.37): u* = V-y{J8 = 10.19
✓0.0388
- -
8
= 0.709 fps
than for the rough pipe. In contrast, the velocity profile in laminar flow is inde- i Finally, from Eq. (8.40),
pendent of pipe roughness.
As we have now derived a theoretical equation for the velocity profile for !i U2in = Umax - 5.76(0.709)1ogf = 12.85 - 1.948 = 10.90 fps ANS
turbulent flow in circular pipes, we can also derive equations for the kinetic-
energy- and momentum-correction factors (Secs. 5.1 and 6.3) using mean veloc- Note: umaxlV = 12.85/10.19 = 1.261. If the flow had been laminar the velocity
ities. Respectively, these equations are 9 ; profile would have been parabolic and umaxlV would have been 2 (Sec. 8.7).
~
~

a - 1 + 2.7f (8.45a)

f3 1. + 0.98f (8.45b)
EXERCISES

8 Although 8.11.1 In a 1.00-m-diameter pipe velocities are measured as 5.35 mis on the centerline
the preceding theory agrees very well with experimental data, it is not
and 4.91 mis at r = 70 mm. Approximately what is the flow rate?
absolutely correct throughout the entire range from the axis to the pipe wall, and
some slight shifts in the numerical constants could improve agreement with test 8.11.2 For turbulent flow in a circular pipe, find r/r0 at the radial distance from the
data. Thus in Eqs. (8.43) and (8.44) we can replace the 1.326 by 1.44, and in centerline where the mean velocity occurs.
Eq. (8.44) although many writers use 2 instead of 2.04, a better practical value seems 8.11.3 Oil (s = 0.92) with a viscosity of0.00065 lb·seclft2 flows at a rate of 6 cfs
to be 2.15.
9 through a 4-in-diameter pipe having f = 0.040. Find the friction head loss.
L. F. Moody, Some Pipe Characteristics of Engineering Interest, Houille Blanche,
Determine the shear stress at the pipe wall and the velocity at 1.5 in from the
May-June, 1950. centerline.
280 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
8.12 PIPE ROUGHNESS
Unfortunately, there is as yet no scientific way of measuring or specifying the
roughness of commercial pipes. Several experimenters have worked with pipes
with artificial roughness produced by various means so that the roughness could
be measured and described by geometric factors, aI?-d they have proved that the
friction depends not only on the size and shape of the projections, but also on
their distribution or spacing. We have much more to do before we completely
solve this problem.
Noteworthy efforts in this direction were made in 1933 by a German engi-
neer, J. Nikuradse, a student of Prandtl's. He coated several different sizes of
pipe with sand grains that he had sorted by sieving into different grain sizes of
reasonably uniform diameters. Let us represent the diameters of the sand grains
bye, which is known as the absolute roughness. In Sec. 8.4 dimensional analysis
of pipe flow showed that for a smooth-walled pipe the friction factor f is a func-
tion of Reynolds number. A more general approach, including e as a parameter,
reveals that f = </>(R, e/D). The term e/D is known as the relative roughness.
In his experimental work Nikuradse had values pf e/D ranging from 0.000985
to 0.0333.
In the case of artificial roughness such as this, the roughness is uniform,
whereas in commercial pipes it is irregular both in size and in distribution. How-
ever, we may describe the irregular roughness of commercial pipe by a single e
value, if we understand this means that the pipe has the same value off at a high
Reynolds number that it would have for a smooth pipe coated with sand grains
of uniform size e. This "equivalent" grain size e must be close to the mean size
of the irregular roughness elements.
Investigators working with pipes have found that if the thickness of the vis-
cous sublayer 8v > e (i.e., eu*/v < 5), the viscous sublayer completely submerges
the effect of e (see Sec. 8.10). Prandtl (Secs. 8.8 and 8.11), using information
from Eq. (8.40) and data from Nikuradse's experiments, developed an equation
for the friction factor for such a case:
Smooth-pipe flow: 1
Vf = 2log
(RV[)
2.51 (8.46)
8.12 Pipe Roughness 281
Because of the way that f appears in two places in Eq. (8.46), it is implicit
inf (see Appendix B) and hard to solve; we must use either iteration or a graph
of fversus R. However, as suggested by Colebrook, 10 we can approximate it by
the explicit equation
Smooth-pipe flow: (8.47)
which differs from Eq. (8.35) by less than ±1.5% for 4000 ~ R ~ 108•
Blasius 11 has shown that for Reynolds numbers between 3000 and 105 we
can approximately express the friction factor for a smooth pipe as
Blasius, smooth pipe,
3000 ~ R ~ 105: (8.48)
Sometimes we can use this very conveniently to simplify equations. Blasius also
found that over the same range of Reynolds numbers, the velocity profile in a
smooth pipe closely agrees with the expression
Blasius, smooth pipe,
3000 ~ R ~ 105: (8.49)
where y = r0 - r, the distance from the pipe wall. We call this equation the
seventh-root law for turbulent-velocity distribution. Though it is not absolutely
accurate, it is useful because it is easy to work with mathematically. At Reynolds
numbers above 105 we must use an exponent somewhat smaller than ~ to give
good results.
At high Reynolds numbers 8v becomes much smaller, and the roughness
elements protrude through the viscous sublayer as in Fig. 8.9b. If 8v < t;ie (i.e.,
eu,/v > 70), investigators have found that the flow behaves asfully-rough-pipe
flow, i.e., the friction factor is independent of the Reynolds number. For such a
case von Karman found that we can express the friction factor as

Fully-rough-pipe _1_ = 21 (3.7)

flow Umin): vJ og e/D (8.50)
This equation applies to turbulent flow in any pipe as long as 8v > e; when this
condition prevails, the flow is known as smooth-pipe flow. Since e does not ap-
The values off from this equation correspond to the right-hand side of the
pear in the equation, it is as though e = 0. Users have found the equation to be
Moody chart (Fig. 8.11), where the curves become horizontal. We sometimes
reliable for smooth pipes for all values of Rover 4000. For such pipes, i.e., drawn
refer to these values as fmin·
tubing, brass, glass, etc., we can extrapolate it with confidence for values of R far
beyond any present experimental values because it is functionally correct, as-
10
suming the wall surface is so smooth that the effects of the projections do not C. F. Colebrook, Turbulent Flow in Pipes, with Particular Reference to the Transition
pierce the viscous sublayer, which becomes increasingly thinner with increasing Region between the Smooth and Rough Pipe Laws, J. Inst. Civil Engrs. (London),
R. That this is so is evident from the fact that the formula yields a value off= 0 Vol. 11, February 1939.
11
for R = co. This agrees with the facts, because R is infinite for a fluid of zero vis- H. Blasius, Das Ahnlichkeitsgesetz bei Reibungsvorgangen in Fltissigkeiten, Forsch
cosity, and for such a case f must be zero. Gebiete lngenieunv. , Vol. 131, 1913.
282 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits 8.13 Chart for Friction Factor 283
In the interval where e > ov > ,he (i.e., 5 < eu*/v < 70) neither smooth
12
flow Eq. (8.46) nor fully rough flow Eq. (8.50) applies. In 1939 Colebrook com- a/a 'ssau46noJ aAffe1al:J
bined Eqs. (8.46) and (8.50) to yield CX) <D
0 0 0 0 U') 0 <DCX) "<t C\J r-00

Turbulent flow, all 1

vJ
(e/D
= -2log 3.7 + Rv]
2.51) (8.51)
0
0
LO "<t"
0
0
(')
0
0
C\I
0
0
0
0
r-00
00 0
oo o
0
0
0
0
0
0
000
000
000
pipes (Colebrook): 0
0 ...; 1.-- ' :-· .. CX)
0 ·· I ·· -·--· -· ,-- ·-- -- ··- ___ _.
ci 1-- / (0

Besides providing a good approximation to conditions in the intermediate 0

0
l·-
_______: ____ __ , _I _ I
/ U')

"<t
range, for e = 0 the Colebrook equation reduces to the smooth-pipe equation 0
0 '. ;1
- / -

0 ~ -- ·-f------f-:---- -- / - C'?
(8.46), and for large R it reduces to the fully-rough-pipe equation (8.50). Thus it 0
"<t' I
I

applies to all turbulent flow conditions. Values of friction factor f that it predicts
' '. I
C\J
0 : :! /
0 I
are generally accurate to within 10-15% of experimental data. This equation is 0
C\J 0 I
I
,...
so useful that engineers have long used it as the accepted design formula for tur- 0
0
0

~
0 0 - / - ----:-· CX)
bulent flow; however, it has one major disadvantage. Like Eq. (8.46), it is im- (/)
0
0
- --- I ·-- ---~---- ---- C\I

plicit inf (Appendix B), which makes it inconvenient to use to manually evalu- .e- T"""
~ <D
U')
E
~
.S E o
ate f. More recently, in 1983 Haaland 13 combined Eqs. (8.47) and (8.50) to -2:-
C
·-
O
0
--- "<t
~
·u 0
0 -2:- "<t" (l) . I
C'? E
provide another approximation, 0
Q) "<t' ·u (.) ' ~
> 0
Q)
0
0
-s
a5 -- +-! C\I E
X > ~
-1 -l.8log [(e/D)1.1
1 0
Turbulent flow, all ...a ! Q
+ -6.9]
0
= - (8.52)
(/)
Q) C\J
X
2 ' 'f:, 0
pipes (Haaland): vJ 3.7 R
.c
(.)
.S 0
E o
E
-------
8 ---
----,----- a> --
__ ,______ Q) CX) g
0 .S - ... ,..... - ci - --
.S (0 ~
Q) Q) E" U') ,:t=
- - 0 - ·--
which has the advantage of being expllcit inf; it has the same asymptotic behav- Q) 0 Q)
0 0 "<t ~
ior as Eq. (8.51), from which it differs by less than ±1.5% for 4000 ~ R ~ 10 •
8 co E
E 0
co 0
co
'.2-
"<t"
C'? .e-ui
'.2- "<t'
C\J ~
LL 0 0
0 U) 0 ~
C\I
<D 0
EXERCISES C\J 9.
°'
Lt)

8.12.1 Using the implicit equation (8.46), the approximate equation (8.47), and Blasius' °'
Q) Q) 0
0
0 ~
$:-
Q
equation (8.48), solve for the smooth-pipe friction factor fusing Reynolds
numbers of (a) 4000, (b) 20,000, and (c) 105 • (d) For which of these three values
°'
3
.E
0

co °'
3
.2 0
<D
ex:,

~ II
~
do the equations show the most variation inf? s:-
<D
s:- 0
"<t"
"<t
Q)
"
8,
"<t' 8, C'? ..0
E
8.12.2 Substitute into Eq. (8.51) the given and computed data of Sample Prob. 8.5a. 0 0 C\I ~
0 (/) C\J
How well does the right-hand side of the equation agree with the left-hand side? (/)
C\J
Q)
:::,
(/)
-0
Q)
:::, 0
8.12.3 Repeat Exer. 8.12.2 using Eq. (8.52). ci5 ~ 0
C
>,
> co £
CX)
(0
<D U')
<D
8.13 CHART FOR FRICTION FACTOR 0 "<t"
"<t
"<t'
The preceding equations for f have been very inconvenient to use in a number 0
of circumstances, which we will discuss further in coming sections, and this C\I
C\I

was especially true before Haaland's equation appeared. The inconvenience was 0
largely overcome by reading numerical values from a chart (Fig. 8.11), prepared
0

12 See Footnote 10. 0

13 S. E. Haaland, Simple and Explicit Formulas for the Friction Factor in Turbulent Pipe 2z/v1(a/7) ·
Flow, J Fluids Eng., Vol. 105, March 1983. j =j JOPBJ UO!P!J:J
lf
lffiil: Programmed computing aids (Appendix C) could help solve problems marked
with this icon.
 8.14 Single-Pipe Flow: Solution Basics 285
284 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits

by Moodyl--l in 1944. The chart, often called the Moody diagram,1 is based on
5
8.14 SINGLE-PIPE FLOW: SOLUTION BASICS
the best information available, and was plotted with the aid of Eqs. (8.29) and The methods of _solut_ion we will use for single-pipe flow are extensions of the
(8.51). All the quantities involved are dimensionless, so the chart may be used method summar~zed m Sec. 5.14. We recommend that yot,1, .review that section
for both BG and SI unit systems. For convenience, BG values of DV ( diameter pow, to?ether w1_th Secs. 5.11 and 5.13, which discuss the energy line and the
tinfos ·velocity) for water at 60°F and similar SI values for water at l5 °C are hydrauhc grade hne.
gi\!e'rfatr·oss the 'tbp of the chart to save the need to compute Reynolds number
for those common cases. Governing Equations
The Moody chart, and the various flow conditions that it represents, di-
vides into four zones: the laminar-flow zone; a critical zone where values are Four simultaneous _equations govern flow at a point in a single pipe. Three of
uncertain because the flow might be either laminar or turbulent; a transition these are the equat10ns of continuity, energy loss, and the Reynolds number:
wne, where f is a function of both Reynolds number and relative pipe rough-
ness; and a zone of complete turbulence (fully-rough-pipe flow), where the value
i -
1 -
4Q
(4.7); hr= f--
L v2 (8.13); (8.1)
nD 2 D 2g V
off is independent of Reynolds number and depends solely upon the relative
roughness, e/D. If the flow is turbulent, as is most common, the fourth is the Colebrook Eq. (8.51)
There is no sharp line of demarcation between the transition zone and the or the H~alan~ Eq. (8.~2) _for the friction factor, as just presented in Sec. 8.12; if
zone of complete turbulence. The dashed line of Fig. 8.11 that separates the two the flow 1s lammar, which 1s more rare, the friction factor is instead given by J =
zones was suggested by R. J. S. Pigott; the equation of this line is R = 3500/ 61/R: Eq. (8.~9). T~e f?ur unknowns are Hsually f, R, D or hr, and Q or V. If the
(e/D). On the right-hand side of the chart the given values of e/D correspond pipelme is umform m size, shape, and roughness, then these equations will yield
to the curves and not to the grid. Note how their spacing varies. The lowest of the same results at all points along it.
the curves in the transition zone is the smooth-pipe ( e = 0) curve given by In some c_as~s we will need hf as a known quantity, and if it is not given we
Eqs. (8.47) and (8.48); notice how many of the other curves blend asymptotically can often obtam 1t by rearranging energy Eq. (5.28) to first find the total head
into the smooth-pipe curve. loss hL as follows:

hL = H1 - H2 = (r + z + v2) - (r + z + v2)
'}' 2g L '}' 2g 2

EXERCISES ~p 1/42- v:2

8.13.1 Oil (s = 0.90) with viscosity 1.6 x 10- 4 lb·sec/ft2 flows in a 4-in-diameter
welded-steel pipe (see Table 8.1) at 0.25 cfs. What is the friction head loss per
foot of pipe?
8.13.2 Oil (s = 0.94) with viscosity 0.0096 N·s/m2 flows in a 90-mm-diameter welded-
steel pipe (see Table 8.1) at 7.2 Lis. What is the friction head loss per meter of
pipe? ·
8.13.3 Water at 20°C flows through a 150-mm-diameter pipe with e = 0.015 mm.
(a) If the mean velocity is 5 m/s, what is the nominal thickness 8v of
the viscous sublayer? (b) What will 8v be if the velocity is increased
to 6.2 mis? ·
8.13.4 A straight, new 48-in-diameter asphalted cast-iron pipe (see Table 8.1) 700 ft
long carries 78°F water at an average velocity of 12 fps. (a) Using the value off
as determined from Fig. 8.11, find the shear force on the pipe. (b) What will be
the shear force if the average velocity is reduced to 4.2 fps?
14 L.F. Moody, Friction Factors for Pipe Flows, ASME Trans., Vol. 66, 1944. Lewis F.
Moody (1880-1953), an eminent American engineer and professor, also contributed
greatly to our understanding of similitude and cavitation as applied to hydraulic
machinery (Secs. 15.4, 15.10, 16.9, and 16.12).
15 Fig. 8.11 is also called a Stanton diagram, because Stanton first proposed such
a plot.
_ + ~z + 1 2
'Y 2g
If, for e~ample, points 1 and 2 are in contact with the atmosphere p = p 2 =
~p = O; if they are also on reservoir water surfaces V1 = V2 = O so that
If the_Y are_m_ · 1·me o f constant diameter, ½ =' Vz. As we noted
· a pipe ' L
= A
LJZ •
h
in Sec. 5.3,
the pipe fnct10n head loss hr must be eit~er equal to or less than hL; the differ-
ence re~ults fro1:1 w_hat are known as mmor losses, which are often negligible,
and which we will discuss more fully in Secs. 8.20-8.26.
Solutio~ of Special Cases
L~minar Flow.. F?r ia11:i~ar flo:V (see Fig. 8.11) we know from Eq. (8.29) that
f - 64/R. Substit~tmg this mto p1pe:friction Eq. (8.13) to eliminate f, and using
Eq._ (8.1) fo~ R, yields Eq. (8.28). This equation is explicit (Appendix B) in each
vanable of mterest, hr, D, and V, so we can solve directly for. any one of them.
We mu~t ~emembe~ to check the validity of the solution we obtain by confirming
that R 1s m the lammar range (normally < 2000, see Sec. 8.2).
Smooth-Pipe Flow._ When e is so small that the roughness elements do not ex-
tend ~hrough the viscous sublayer or zone (i.e., e < 8v, see Sec. 8.10), so that
the pipe beh~v~s as a smooth pipe (Fig. 8.9a) even though most of the flow is
turbulent, this 1s represented by the lowest curve on Fig. 8.11, for which e = O.
286 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits 8.15 Single-Pipe Flow: Solution by Trials 287
From Sec. 8.12 we know we may express f along this curve by the Blasius equa- In engineering applications, flows in any of these three special cases
tion (8.48) provided 3000 ::::; R ::::; 108 . Combining Eq. (8.48) with pipe-friction occur relatively rarely. In Fig. 8.11 we see that the only flow region not covered
Eq. (8.13) to eliminate f, and using Eq. (8.1) for R, yields by these three cases is the transition zone. In addition to the fact that most en-
gineering flows occur in the transition zone, flow solutions for this zone are
· LVu 5
Smooth pipe, 5 hf = 0.l580zi·2s___ (8.53) more involved. We will discuss various ways to so]ve such flows, and flows in
3000 ::::; R ::::; 10 : gD1.2) which the regimen (zone) is initially unknown, in Secs. 8.15-8.19 and 8.27.
This equation has the advantage of being explicit in each variable of interest, hf,
D , or V We must remember to check the validity of the solution it yields by con-
firming that R is in the required Blasius range. 8.15 SINGLE-PIPE FLOW: SOLUTION BY TRIALS
Fully-Rough-Pipe Flow. When e is so large that the roughness elements As we have already noted in Secs. 8.12-8.14, often some of the equations govern-
extend through both the viscous zone and the transition zone (i.e., e > l4ov; ing pipe flow are implicit in form (Appendix B), and therefore they do not readily
see Fig. 8.8), so that there is complete turbulence and the flow is fully rough permit direct solution. For example, although Haaland's equation (8.52) is explicit
(Secs. 8.10 and 8.13), this is represented by the horizontal lines at the right of inf, it is still implicit in Vand D, since these are involved in both Rand inf through
Fig. 8.11, where R > 3500/(e/D). Then/= hnin = constant is given by Eq. (8.50), Eq. (8.13). The classical way to overcome this difficulty has been to use trial-and-
which when rearranged gives error procedures in combination with the Moody chart described in Sec. 8.13.
2 With the more recent availability of the Haaland equation, we can now also per-
Fully rough 3.7)]- (8.54) form these procedures using equations alone. We will expfore both methods in
pipe:
f = fmin = [ 2log ( e/D this section, including applications to different types of problems.
To solve any problem (other than laminar flow or smooth-pipe flow) using
Substituting this into pipe-friction Eq. (8.13) to eliminate f, yields the friction factor f, we will require a value of e, and we can obtain this from
Fully rough 1 L V2 Table 8.1. As the ratio of e/D is dimensionless, we may use any units, provided
2
(8.55) they are the same for both e and D. Note also that we should use exact values
pipe (fi11in): hf = [ (3.7)] D 2g
2log e/D of the internal- pipe diameter. Exact diameters often differ from the nominal
sizes, particularly for smaller pipes. If the pipe is not of circular cross section, we
We notice that for fully-rough-pipe flow hf is directly proportional to V and is
2 can replace D in all instances by 4Rh as we noted in Sec. 8.6.
independent of R. Equation (8.55) is explicit in hf and V, but implicit in D (see With regard to the values of e, note that those given in Table 8.1 are for
Appendix B). The solution of implicit equations is discussed in Secs. 8.15-8.17 new, clean pipes, and even in such cases there may be considerable variation in
and 8.27, and in Appendixes C and D. Values of e are given in Table 8.1. We the values. Consequently, in practical cases, the value off may be in error by
must check the validity of a solution we obtain from this equation by confirming ±5% for smooth pipes and by ±10% for rough ones. For old pipes, values of e
that R is sufficiently large. may be much higher, but there is much variation in the degree with which pipe
roughness increases with age, since so much depends on the nature of the fluid
transported. In addition, deposits in small pipes materially reduce the internal
TABLE 8.1 Values of absolute roughness e for new commercial pipes diameter. The effect of the roughness of pipe joints may also increase the value
Feet Millimeters off substantially. So we must use judgment in estimating a value of e, and con-
sequently off.
Glass, plastic (smooth) 0.0 0.0 Most engineering problems fall within the transition zone (see Fig. 8.11).
Drawn tubing, brass, lead, copper, centrifugally spun 0.000005 0.0015 When solving single-pipe flow problems in this zone, where conditions depend
cement, bituminous lining, transite on both e/D and the Reynolds number, we need a definite value of R. But when
Commercial steel, wrought iron, welded-steel pipe 0.00015 0.046 V or Dis unknown, so is R. The solution procedure varies with the type of prob-
Asphalt-dipped cast iron · 0.0004 0.12
lem, and we can categorize most single-pipe flow problems into one of the fol-
Galvanized iron 0.0005 0.15
0.25
lowing three types:
Cast iron, average 0.00085
Wood stave 0.0006-0.003 0.18-0.9
0.3-3· Type Find Given
Concrete 0.001-0.01
Riveted steel 0.003-0.03 0.9-9 1. Head-loss problem D, Q or V, and g, L, e,
ht JI
2. Discharge problem Q or V D,h1, and g, L, e, JI
e e in feet e in feet einmm 3. Sizing prohlem D Q,h1, and g, L, e, JI
Note: D = D.m f eet 12x .. h Din mm
D m me es
288 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits 8.15 Single-Pfpe Flow: Solution by Trials 289
Here g is the gravitational acceleration, L is the pipe length, Q is the flow rate, Note th at in all cases we should assume an f value whenever possible,
and vis the kinematic viscosity. Of course, we may knowµ, and p instead of v , in rather than some other vari able . This is because /varies the least, so trial calcu-
which case we may obtain v from Eq. (2.11). lations are less sensitive to its changes, and as a result they converge faster.
Type I problems are the simplest. If Q is given, we can easily obtain V from Sample Probs. 8.4-8.6 illustrate solutions to Type 1, 2, and 3 problems; the
continuity Eq. (4.7). Then we can easily obtain the Reynolds riumber and e/D from last two of th ese use trial and error.
the given data. Next we enter Fig. 8.11 vertically with Rand along a curve (possi-
bly interpolated) for e/D from the right, to identify an operating point for which we
r·~.,.._l'.,&;
can read off the friction factor f horizontally to the left. Alternatively, we can com- ...,,....U"'t.'1>11..~· . . . "'VIX• ,.,_.)..~#';' fo.l"'" sa ---=-:........,,.,,, .. ~ ~ ~ - : . , ; . : r•
. 1 , ¥ ~ ~· -Q~lellt;.i.. ·-e..:r.-....-..~:·:!I.: " ... "

)
pute /from Eq. (8.52) or by trial and error from Colebrook Eq. (8.51). With this 1 SAMPLE PROBLEM 8.4 A 20-in-diameter galyanized iron pipe 2 miles long
value off, we can directly compute the friction head loss h1 from Eq. (8.13). carries 4 cfs of water at 60°F. Find the friction head loss: (a) using Fig. 8.11 and
In Type 2 and Type 3 problems, because either V or D is unknown, the the Reynolds number; (b) using Fig. 8.11 and its D"V scale; (c) using only a basic
Reynolds number R is not known at the outset and so a direct solution is not scientific calculator,17 without Fig. 8.11.
possible. However, we notice in Fig. 8.11 that the value off changes very slowly
Solution . .
with large changes in R. So we can usually solve the problem quite effectively by
assuming an initial value off, and then obtaining the final solution by successive This is a Type 1 problem, to find h_r- From Table 8.1 for galvanized iron:
trials (trial and error; see Sample Prob. 3.5). If D is known (Type 2) , the fully- e = 0.0005 ft: so e/D = 0.0005(12)/20 == 0.0003
rough-pipe f value Umin) given on the right-hand side of Fig. 8.11 or by Eq. (8.54)
provides a good starting point. If Dis not known (Type 3), a value near the mid- L = 2 mi(5280 ft/mi) - 10,560 ft
dle of the /range on Fig. 8.11 , such as 0.03, makes a good start. Each succeeding
trial is started with the f value obtained from the previous trial. The value off 4Q 4(4)
can be considered close enough when the first three significant figures of the re-
Eq. (4.7): v-- -ftD -
2 - n(20/12) 2
1.833 fps
quired answer (Q, V, or D) no longer change. Note that if we assume values of
some other variable besides f , convergence is usually much slower. Table Al for water at 60°F: v = 1.217 x 10- 5 ft 2/sec
For Type 2 problems (to find V or Q), we rearrange Eq. (8.13) into the
form V = K/Vj, where K = -V2gDh/L is known. Assuming an f (as just dis- DV (20/12)1.833
cussed) therefore yields a V, which enables us to calculate R and enter the R = 2.51 x 105 (> Rcrit, i.e.•, flow is turbulent)
l) 1.217 X 10 - 5 .
diagram or to use Eq. (8.51) or (8.52) to obtain an improved value of J. If this is
different from the assumed f, we must repeat the procedure assuming the just- (a) Enter Fig. 8~11 at the right-hand side with e/D = 0.0003, by interpolating
obtained value, and successively repeat it until the two values converge. This between 0.0002 and 0.0004; note that the e/D spacing varies. Follow this
usually only requires two or three trials, by which time all the values are correct, (unplotted) e/D curve to the left until it crosses a vertical line at R = 2.51 x 105
including the required value of V. (c;FZution: this is between 105 and 106 ). For this operating point, reading hori-
For Type 3 problems, since the D to be found is unknown, neither e/D nor zontally to the left, f = 0.0172.
Rare known initially. We substitute V = 4Q/(nD 2 ) into Eq. (8.13) and rearrange
it to obtain D = (JK)115 , where K = 8LQ3/(n2gh1) is known. Then assuming an f L V2 . . (10,560) _l.833 2
Eq. (8.1)): ht = f D 2g = O.Ol 72 (20/12) 2(32.2) 5.69 ft ANS
( as previously discussed) yields a D, which enables us to proceed with repetition
in a manner similar to that used for Type 2.
Experience has shown that new users of the Moody diagram frequently (b) D"V = 20(1.833) = 3(j.7. Find this value on the scale across the top of
misread it. This most probably occurs because none of the scales are linear, and Fig. 8.11; note that this scale is varying. Find where the (interpolated) curve for
because the intervals between grid lines and chart curves keep changing. So take e/D = 0.0003 crosses the vertical line at D"V = 36.7. From this point, read hori-
special care in reading the chart, and confirm any interpolated values by com- zontally to the left, to find/= 0.0172. Compute h_r as for part (a).
paring them with nearby grid values or curves in both directions. 16 Note: From the operating point on Fig. 8.11 , we see that flow conditions are in
the transition zone of turbulent flow, which is typical.

16
Charts involving the same functional relations have been plotted with different
coordinates from those in Fig. 8.11 and may be more convenient for certain specific
purposes, but we believe that the form shown is best both for purposes of instruction t 7 We define a basic scientific calculator as one that is not programmable and does not
and for general use. have automatic equation solving capabilities .
290 CHAPTER 8: Steady Incompressible Flow in. Pressure Conduits 8.15. Single-Pipe Flow: Solution by Trials 291
1 0.0003)1.ll 6.9 ] (b) Eq. (8.54) for e/D = 0.000092: fmin = 0.01179. Calculate V and Ras m (a), ,
(c) Eq. (8.41): -l.8log [( 3:/ + 2.51 x 10s = 7.65 then obtain an improved f from Eq. (8.52). Use the obtained /for the next trial.
vJ Tabulating all the trials:
from which f = 0.01709
Tryf V,m/s R Obtained!
10,560 (1.883) 2
ANS 0.01179 2.23 1.114 X 106 0.01309 Try again
Eq. (8.10): ht = O.Ol 7 09 (20/l 2 ) 2 ( 32 _2 ) = 5.65 ft

Note: This differs from answer (a) by only 0.04 ft, or 0.70%.
~
I 0.01309

Q =
2.12

AV =
1.057 X 106

(n/4)(0.5)22.12 = 0.416 m 3/s

0.01315

ANS
Converged!

'1' ....... t; -: .... ....;.:J( ,;..;1·:,c-.1" ~· ",'+

...... t.,r...-.,...~.,..,_...,,~. ~._. .-ir~.::r_._.......

- ~~<\)I

""'11•
•

!r.,
..; ,;-;

~<I:
,-,.:

,,..,_
v-.-:.. -,,...:-1,~'41,.."'!:;'

-~... .1.--i:•~"I:
Jlo:W':;\~-,11'!"... -t.~,•.\"': ....

:S.~""~ •• ..----....ci·
-.u ...... -,.'="""'-"'-.1 ·-

'...xi-....::it'J'- ...~·.:1~~....-;:"lo·.'(-...,...
"·""'

::~.i L_ - -~-
i
SAMPLE PROBLEM Water at 20°C flows in a 500-mm-diameter welded- ~
8.5 r~~,.
steel pipe. If the friction loss gradient is 0.006, determine the flow rate: (a) using ~
~~ " . r . ' \ ' . " ~ ~ i A " " . ~ . r J l e " \ . ~ . f " - " . ~ ~ ~ . T , - , ' a , ~....».Y»~s~.:.....'"";:;.t;.~.u,,o....,,u··,..""1~T.. ~~I'~~-~~~ ,»1:/K::f:.flr.~

~
Fig. 8.11; (b) using only a basic scientific calculator, 17 without Fig. 8.11. SAMPLE PROBLEM 8.6 A plvanized iron pipe 18,000 ft long must. convey
ethyl alcohol (11 = 2.3 x 10- 5 ft /sec) at a rate of 135 gpm. If the friction head loss
Solution must be 215 ft, determine the pipe size theoretically required: (a) using Fig. 8.11;
This is a Type 2 problem, to find Q. (b) using only a basic scientific calculator, without Fig. 8.11.
Table 8.1 for welded steel: e = 0.046 mm; e/D = 0.046/500 = 0.000092
Solution
Table A.1 at 20°C: 11 = l.003 x 10- m 2/s; 6
htfL = 0.006 is given This is a Type 3 problem, to find D.

_ ht _ j_ V 2 • fV
2 Q = 135 gpm(2.23 cfs/1000 gpm) = 0.301 cfs
Eq. (8.14): S - L - D g, 1.e., 0.006 = O.S( 2 )9 _81
2 Table 8.1 for galvanized iron: e = 0.0005 ft.
from which = 0.243/j112.
V e 0.0005
D D
(a) Fig. 8.11 for e/D = 0.000092: fmin = 0.0118.

Try f = .0.0118. Then V = 0.243/(0.0118) 112 = 2.23 mis. Eq. (4.7):

V = 4Q = 4(0.301) = 0.383
nD 2 nD 2 D2
DV 0.5(2.23) _ .
Eq. (8.1): R = - = - 1.114 x 106 ( turbulent flow) R = DV = D (0.383) 16,590
11 l.003 X 10- 6 Eq. (8.1):
11 2.3 x 10- 5 . D 2 D
Figure 8.11 with e/D = 0.000092 and R = 1.114 x 106: f = 0.0131. Assumed and 2
' 18,000 (0.383) . 1/ 5
obtainedfvalues are different, so we must try again; use the obtainedffor the Eq. (8.13): 21 5 = ht = f D( . ) IJ2 , from which D = 0.718f
2 32 2
next trial. Tabulating all the trials:
(a) Start by assuming a mid-range value off
Try f V,mls R Obtained!

0.0117 2.25 1.120 X 106 0.0131 Try again Tryf D, ft e/D R Chartf
6 Converged!
0.0131 2.12 1.059 X 10 0.0131
0.0300 0.356 0.001404 4.68 X 104 0.0253 Try again
4
The f values now agree, so we have the true operating poin_t. Convergence is 0.0253 0.344 0.001453 4.84 X 10 0.0253 Converged!
rapid!
3
Q = AV .= (n/4)D 2 V = (n/4)(0.5)22.12 = 0.416 m /s ANS Values off now agree, so \Ve have the true operating point. Convergence is rapid!

Caution: Take great care to read Fig. 8.11 correctly. D = 0.344 ft = 4.13 in ANS
292 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits 8.16 Single-Pipe Flow: Direct Solutions 293
(b) · Start by assuming a mid-range value off. S:alcu~ate Vand R as befo~e, then 8.16 SINGLE-PIPE FLOW: DIRECT SOLUTIONS 18
obtain an improved f from Eq. (8.52) . .Tabulatmg tlus and subs~quent tnals:
With time and experience, engineers working on pipe-flow problems have devel-
oped some ingeneous ways to obtain direct solutions. These have the advantage
Try f D, ft e/D R Eq. (8.41)/ of avoiding the tedious trial and error described in Sec. 8.15, and in addition they
are more precise and reliable than reading off a chart that may easily be misread.
0.0300 0.356 0.001404 46,800 0.02501 Try again For people who do not have access to advanced programmable calculators
0.0250 0.343 0.001456 48.500 0.02504 Converged! or computers with mathematics software (Sec. 8.17), ways have been found to
solve each of the three types of pipe-flow problems mentioned in Sec. 8.15 with-
out iteration, so that we can evaluate them directly on a basic scientific calcula-
. D = 0.343 ft 4.1.2 in ANS
tor.19 This of course requires an explicit equation in the sought quantity (Ap-
pendix B). We describe these methods further below.

The Head-Loss Problem (Type 1). The Haaland equation (8.52) makes
As noted earlier, in practice there can be a considerable _a mount of uncer- possible a direct computation off, from which we can find h1 using Eq. (8.13).
tainty in the size of the absolute roughness e. So it is important to have some Section 8.15 and Sample Prob. 8.4 explain this more fully. For a faster solution
idea what effects changes in e will have on h1 , Q, and D. F:01?1 Eqs. (8._13) and we may eliminate f between Eqs. (8.13) and (8.52). ·
( 4.7), it follows that hf ex f, Q ex F 112 , and D ex (1 15 . _The vanatrnn off w1.th ewe
may see in Fig. 8.11 or Eq. (8.51). As an example, m Sample Prob. 8.4, if e had The Discharge Problem (Type 2). We can obtain a single equation for the
been 20% larger then the hf would have been 2.3% larger. This change in the velocity ( or discharge) as follows. First we rearrange Eq. (8.13) to obtain
head loss would be larger for larger f (or larger e/D), and vice versa. Changes in
Q and D will be smaller than those in ht _l__
vJ -
v) L
2gDhf
We can avoid the repetitious labor required for the trial-and-error proce-
dures just described by using direct methods and automated methods. We will Then, when we substitute this and R = DV/v into the Colebrook equation (8.51)
review these alternative solution methods in Secs. 8.16 and 8.17. and rearrange, we get

Turbulent flow, _ _ ~ (e/D + 2.5lv ✓ L )

Type 2: V -
2
\J ~ log 3.7 D 2gDhf (8.56a)
EXERCISES
This equation is_explicit in V, which fortunately results because the Vs from R
8.15.1 Compute the friction head per 100 f.t of 3-in-diameter pipe for a Reynolds and vJ cancel in the last term of Eq. (8.51) [the same substitution into the
number of 50,000 if (a) the flow is laminar (achievable with great.care);
Haaland equation (8.52) does not produce such a desirable result]. Having this
(b) the flow is turbulent in a smooth pipe; (c) the flow is turbulent in a rough
pipe with e/D = 0.05. Consider two situations: one where the ~ui~ is _
explicit equation available will save much repetitive work in solving more in-
70°F water, the other where the fluid is SAE 10 (Western lubncatmg) 011 at volved problems, like those of branching pipes (Sec. 8.26). If Q is required rather
150°F. than V, we can use continuity equation (4.7) to eliminate V, and Eq. (8.56a)
becomes
8.15.2 California crude oil, warmed until its kinematic viscosity is 0.0004 ft 2/sec
and its specific weight i's 53.45 lb/ft3, is pumped through a 3-in pipe Turbulent flow, 4Q = _ ~ (e/D + 2.Slv✓ L )
(e = 0.001 in); (a) For laminar flow with R = R crit = 2000, what would
be the loss in energy head in psi per 1000 ft of pipe? (b) What would be
Type 2: nb 2 2
\J ~ log 3.7 D 2gDhf (8.56b)

the loss in head per 1000 ft if the velocity were three times the value
in (a)?
8.15.3 Water at 50°F flowing through 80 ft of 4-in-diameter average cast-iron pipe
causes a friction head loss of 0.27 ft. Find the flow rate.
8.15.4 When gasoline with a kinematic viscosity of 5 x 10- 7 m 2/s flows in a
200-mm-diameter smooth pipe, the friction head loss is 0.43 m per 1-00 m.
Find the flow rate.
18
Some users may omit this section because it is not required to understand the
governing equations and their relationships. Note, however, that the methods described
here are of great benefit in Secs. 8.24 (single pipe with minor losses) and 8.26
(branching pipes).
19
We define a basic scientific calculator to be one that is not programmable and does
not have automatic equation solving capabilities.
 8.16 Single-Pipe Flow: Direct Solutions 295
CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
294
Using Eq. (8.56), we can calculate Q or V directly with a basic scientific calcula- SA MPLbE ~RO~LE~ 8.7 Solve Sample Prob. 8.5 without trial and error using
tor. Remember to use Eq. (8.1) to confirm that R is in the turbulent range. If R on 1ya as1c sc1ent1fic calculator. '
turns out to be in the laminar range, i.e. , less than 2000. we must instead find V
Solution
or Q from Eq. (8.28) rearranged. ~his is a Type 2 problem, to find Q. As in Sample Prob. 8.5, D = 0.5 m
, htfL - 0.006, e/D = 0.000092, and v = 1.003 x 10- 6 mt!s. The quantity '
The Sizing Problem (Type 3). We can eliminate iteration by reformulating our

~
equations as follows. Substituting from Eqs. (8.13), (4.7), and (8.1) into the di-
= v'2(9.81)0.5(0.006) = 0.243 mis,
mensionless quantity
2
(D 4Q \5 = 12~g ht Q~
2
N = Rs = h1D2g (nD ) (8.57)
f L 4Q v nD 2 ) n:> L ., ,.:, so in turbulent-flow Eq. (8.56a):
1

indicates that it is independent of the unknown D, so that N 1 is a known quan- ·

tity. We can also transform the relative roughness into another known dimen-
v = _ 2 (0 .243 )log[0.000092 +
3.7
2.51(1.003 x 10 -6)
0.5
_l_l
0 .243 = 2.11 mis
sionless quantity by dividing it by R , i.e.,

N
2
- e: -~(fv) = ~(1):~
2

= :~ (8.58)
Check: R = DV =
V
0.5(2.11)
1.003 X 10- 6
= 1.050 x 106 . This is >

so flow is turbulent, validating the use of Eq. (8.56a ). Finally,

R cnt. = 2000

Now we can use N to eliminate ffrom laminar flow Eq. (8.29) to obtain
1 Q = AV = (n/~)D 2V = 0.25n(0.5) 2 2.11 = 0.414 m 3/s ANS
N1)0.2s
(8.59)
Laminar flow, R = ( --
64
Type 3:
and we can use N and N to eliminate f and e/D from the Colebrook equa- SA MPLbE ~RO~LE~fi1 8.8 Solve Sample Prob. 8.6 without trial and error using
1 2
tion (8.51) for turbulent flow to obtain 0 n 1ya as1c sc1ent1 c calculator.
N2R 2.51 l
R2·5 = -2N 01·S log ( - - + --- R ··
'i) (8.60) Solution
3.7 N~·) r _ This is a Type 3 problem, to find D. As in Sample Prob. 8.6, ht = 215 ft
L - 18,000 ft, Q = 0.301 cfs, e = 0.0005 ft, g = 32.2 ft/sec1, and v = 2.3 x 10-s ft2/sec'.
Here R occurs in three places, so thii; equation is strongly implicit (Appendix B),
3
and Eq. (8.60) appears to be not a very useful result. However, if we plot N 1 versus
Eq. (S.S?): Ni = 128(32.2) ( ~ ) (0.301 ) _ 6 71 21
R we find it collapses the various flow curves on the Moody chart into 0 a very nar- 7r 3 18,000 (2.3 X lQ - 5)5 - · X lQ
row band that is cloSeiy approximated by the formula R = l.43N1·' '. We can sub-
stitute this into tbe right -hand side of Eq. ( 8.60) to create the more useful equation: nev n(0.0005)2.3 x 10-s

Turbulent flow, R2.s = -2No.s log (

N 2No.208
1 4 29
+ - ·188
- (8.61)
0 Eq. (8.58): N2 = -
4Q
=
4(0.301)
3.00 X lQ - 8

1
Type 3: 2.59 N~·
Laminar flow Eq. (8.59): R = (N641)0.2s = 101,300 > R crit = 2000
We see that this reformulation has converted the Colebrook equation into an
explicit form.
Comparison of the two Reynolds numbers from Eqs. (8.59) and (8.61) with Turbulen~-flow Eq. (8.61):
R . = 2000 indicates which equation (laminar or turbulent) is applicable for a
cnt
given Type 3 problem. .
Finally, from Eq. (8.1) combined with Eq. (4.7), we have
R2.s = -2(6.73 x 1021)°-slog ~ 3.00 x 10-s(6.71 x 1021)°-208 +
l. 2.59
4.29
(6.71 X 1021)°'188
l
= 5.13 10 11
R = D: = ~(:;2) :?v =
X

and so, by rearrangement, we can obtain the required diameter D from

~th: tpl ~o~rammed computing aids (Appendix C) could help solve problems marked
4Q
Type 3: D = - (8.62) w1 11s icon.
nvR
 /I

. ·b/e Flow in Pressure Conduits

8.17 Single-Pipe Flow: Automated Solutions 291
CHAPTER 8: Steady I ncomp1 essi
296 calculators and the equation solving software is that at present (2001) the calcu-
. l R - 48 300 Clearly the laminar flow equation is invalid, flow is \ lators can solve only one nonlinear ( and implicit) equation for one unknown
from wh1c l, d R-- 4's 300 Thus from Eq. (8.62), ~

J
(Appendixes Band C.1), whereas a number of software packages like Mathcad
turbulent, an - , · ' ,
and Excel can automatically solve a system of many nonlinear equations in an
4Q 4 3
(0. 0l) = 0.345 ft ANS . equal number of unknowns.
5
D = J[VR = n(2.3 x 1o- )48,300 Next we will discuss ways to use automated equation solvers with each of
. l O00055 ft or 0.16% from the accurate answers m the three principal types of single-pipe flow problem.
Note: This differs by on Y ·
Sample Prob. 8.9. w, ...., , ...... ~_... ....==-"'"- _.,,._..--..-- The Head-Loss Problem (Type 1). While equation solvers are not necessary
"1•'1 •' - - ~ . ~ ........':'"J.~'09
,.....<C'-'J: .......--· ..¥" ~-,u..r:.
for Type 1 problems, we can still use them, as described next for Types 2 and 3.
They are necessary if we want to use the implicit Colebrook equation (8.51).

The Discharge Problem (Type 2). An equation solver on a programmable

EXERCISES . calculator is not necessary for Type 2 problems, because we can solve them
. . l d er1·or using only a basic scientific calculator.
E 8 15 3 without tna an , directly with a basic scientific calculator, as described in Sec. 8.16. However, we
8.16.1 Solve xer. . . . ly a basic scientific calculator.
without trial and error, usmg on can use a programmable calculator for convenience, in. the same way as for
E 8 15
8.16.2 Solve xer. • · 4
Type 3 problems.
When using equation solving software such as Mathcad or Excel to solve
the discharge problem, we enter the four governing equations in their familiar
PIPE FLow· AUTOMATED SOLUTIONS20 forms (Sec. 8.14). We must assign values to the known variables g, v, h, L, e,
8.17 SINGLE- . . lution techniques, automated method~ have and D, and initial estimates for the unknown variables f (0.03 suggested), V, R
With progress m technology and so bl that involve the complicated (> 2000), and Q. The solver then simply solves the four equations (some non-
. 1 . pipe-flow pro ems .
become available for so vmg . . . S 8 12 These methods provide accu- linear, some implicit) for the four unknowns.
equations for ro_ugh nes
s descnbed
.
m
h
ec. · · · s
anual methods described m ecs. ·
8 15
rate and convenient alternatives to t e m The Sizing Problem (Type 3). When using equation solving software, we
and 8.16. . ed e uation solvers have become available follow the same procedure as for the discharge problem just described, with the
In recent years, preprogramm scientific calculators and in some ti one difference that we interchange Q and Don the lists of known and unknown
on certain more advance~ programma a:ka es (see Appendixes C abd D). To variables.
spreadsheet and mathematics so:tware iion !xplicit or implicit (Appendix B) With an equation solver on a programmable calculator, we can solve
use them, we must first enter t e eq~ ' or into a computer file. Such Eq. (8.56b) for D directly, even though it is implicit in D. We must then check
' and in any form, into the calculat_or s m~tmros~rwl11·ch present the equation on the value of R.
"equation wn e , · 1
calculators now often h ave . ·t aper Some of the equation so v-
. f mas we wnte i on p . .h
the screen m the same or . h
ing software packages als~ presen~ :d~t
f on in the same form. In eit er
~f necessary. Then we assign val- ~J:~ fll"r"!'"~-A.:rc•.~~~~~.-...-~.:. n . : ~ ~ -;,.-::'!l:IS"'.:tt,""'~~..,J1/Jd"U>c."'l"'C.,4-:ti~~~.~~,..,~
I
~~a:~lr:-....1"\.M:..,,l!')'llfllt4~~...&.t!lll.~aJ~~~~as

case we can save the equat10:1s, an d . n estimated values to the unknown

ues to each of the known vanables, an ass1g
~~ SAMPLE PROBLEM 8.9 Solve Sample Prob. 8.6 using (a) an equation solver
IBllll f on a programmable scientific calculator; (b) equation solving software.
variables. lves for the unknown variable(s), it
When the calculator or soft_ware dso ocedure to find them, to high I Solution
uses an internally program
med tnal-an -error pr
.
h.
• the fact that once everyt mg is
· This is a Type 3 problem, to find D. As in Sample Prob. 8.6, h = 215 ft, L =
f al convenience 1s ' . t 18,000 ft, Q = 0.301 cfs, e == 0.0005 ft, g = 32.2 ft/~ec 2 , and v = 2.3 x 10- 5 ft2 /sec.
Precision. Of great. •prac iceasy to change w h.1ch of the variables we wish to trea
bl
stored in memory, it i~ very d"ff e between the advanced programma e (a) Using an "equation writer" on the programmable calculator, key in
as the unknowns. An important 1 erenc Eq. (8.56b) and store it permanently, assigning it a name such as EQ856B. Then
activate the equation solving feature (see Appendixes C.1 and D.l), select this
equation from the catalog of equations, and make it the "current" equation
. ·tis not required to understand the (note that it eliminates f, R , and V).
20Some users may omit this ~ection_beca~s: ~ote, however, that in Sec. 8.24 (sing~e Display the menu of variables in the equation, and enter numeric values
governing equations and their rel~t~;~:i9 (branching pipes, series and parallel pipes, for them all (g, v, h1, L, e, Q, D) as above, including an estimate of D (l ft, say).
pipe with minor losses) a!1d Secs. . . f the methods described here.
and pipe networks) we discuss advantages o
 8.18 Empirical Equations for Single-Pipe Flow 299
-,, 'E R 8· Stea(fr Incompressible Flow in Pressure Conduits
298 C IIAl l . • . primary variables of interest (here V, Rh or D, and S = h/L) are based on obser-
vations rather than theory, are known as empirical equations (Appendix B). This
Instruct the solver to solve for D · is in contrast to the Darcy-Weisbach Eq. (8.13), for which we saw in Secs. 8.4 and
D = 0.344 ft ANS 8.5 that these relatiom; were developed from theory, primarily dimensional analy-
Result: sis; the fact that the roughness coefficients were determined experimentally does
·x D 1 oives more information about using solvers on program- not affect the way in which the variable relations were developed.
N ote: A ppen d 1 . o
mablc scientific calculators. . Perhaps the best example of such an empirical equation is that of Hazen and
· , . l E eel or Mathcad (Appendixes Williams, applicable only to the flow of water in pipes larger than 2 in (50 mm)
(b) Using equation solv.1,ng s~{~\ :~ ~,as~~ ;y~~ i; the four governing equations and at velocities less than 10 fps (3 mis), but widely used in the waterworks in-
C.2-C.3), open a file or_ wo~· ~ le te'-n variables and of these we wish to findf, dustry. This formula takes the form
(Sec. 8.14). These equations ITI\O ve '
V, R ,Aand_ D. merical values to the six known variables (g, v, hf, L_, be,l Q),fas BG units: (8.63a)
ss1gn nu . . ' lues to the four unknown vana es, or
before. T~en assign ~st11~a~~O~aOOO D = 0.5. lt is helpful to keep these two
example, f = 0.03, V - 5 , . ' ,. ' , can be labeled with text. The success SI units: V = 0.849CmvR2· 63 S° 54 (8.63b)
sets of assignments on separate_tl_mest, tl~~~ estimated values, so if the solver is
of the procedure may be sen~1. ,1ve o . where Rh (ft or m) is the hydraulic radius (Sec. 8.3), and S = h/L, the energy gra-
unable to find a solution, try different estimates. dient. The advantage of Eq. (8.63) over the standard pipe-friction formula is
-- V = 3 23 ft/s. R = 48,400, D = 0.344 ft
0 0253 ANS that the roughness coefficient Cmv is not a function of the Reynolds number, and
Resu l ts: f •· ,
so we need not make trial solutions. Values of Cmv range from 140 for very
. 1 T e 2 problems in this way. Appen-
Notes: We can als? automat1cal11 y so ;l~mI~olved by Excel and Mathcad. smooth, straight pipe down to 110 for new riveted-steel and vitrified pipe and to
dixes D.2 and D.3 include samp c pro .. 90 or 80 for old and tuberculated pipe.
Another empirical formula, which we will discuss in detail in Sec. 10.2, is
the Manning formula, which is

BG units: V= 1.486 R2;3 5112 (8.64a)

EXERCISES n h
8.17.1 Solve Sample Prob. 8.5 using an equation solver on (a) a programmable
scientific calculator; (b) computer software. . .
SI units: V= .!-R2/351/2
n h (8.64b)
8.17.2 Solve Exer. 8.15.3 using an equation solver on (a) a programmable sc1ent1fic
calculator; (b) computer software. . .
where n is a roughness coefficient, varying from 0.008 for the smoothest brass or
8.17.3 Solve Exer. 8.15.4 using an equation solver on (a) a programmable sc1ent1fic
plastic pipe, to 0.014 for average drainage tile or vitrified sewer pipe, to
calculator; (b) computer software.
0.021-0.030 for corrugated metal, and up to 0.035 for tuberculated cast-iron pipe
(Table 10.1). The Manning formula applies to about the same flow range as does
the Hazen-Williams formula.
EMPIRICAL EQUATIONS FOR SINGLE-PIPE FLOW For some problems, it is more convenient to work with Eqs. (8.63) and
8.18 . . •ven in Secs 8 1-8 17 incorporates the (8.64) i;. the form of expressions for head loss (see Sec. 8.19). Because we can also
!~
The presentation of_ friction l~~s m r:~s far as appli.cation.to Newtonian fluids express the equations in terms of V or Q, depending on which is given or sought,
best knowle~ge availabledonAt d1sms1~tt~dly' however the trial-and-error type of and in BG or SI units, the number of alternative forms is quite large. Because
(s ec. 2 .11) 1s .concerne . b d' with computations
' . for re 1ative
· roug h- each variable occurs only once in the above empirical equations, they are always
solution, especially when ~nc~e:rt:dious when repeated often for similar ~on- explicit regardless of how they are rearranged or which variable is unknown. This
ness and Reynolds number, ec . develo ed simple and convement- gives them their distinct advantage, that we can always solve them directly.
ditions. It is natural, therefor~ that eng1~ee:~ts and ~bservations but limited to Engineers have developed nomographic charts and diagrams for the appli-
to-u~e des~gn formulads: ?ase si;h e;qp::::ns, in which the relations between the cation of Eqs. (8.63) and (8.64). The lack of accuracy that results from using
specific flmds and con itions. . these formulas is not important in the design of water distribution systems, since
we can seldom predict the capacity requirements with high precision. and be-
. . ds (Appendix C) could help solve problems marked cause flows vary considerably throughout the day.
lffl: Programmed computing at
with this icon.
300 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits 8.20 Minor Losses in Turbulent Flow 301
For SI units, we must.replace the 4.66 by 10.29 in the Manning K; recall the same
EXERCISES restrictions on D and V.
8.18.1 When water flows at 2.5 cfs through a 20-in-diameter pipeline, the head loss is - We see that h1 = K Q 2 except when working with the Hazen-Williams coef-
0.0004 ft/ft. Find the value of the Hazen-Williams coefficient. ficient, and we notice that K is a property of the pipe alone. When we need to
solve for discharge, we can of course rearrange Eq. (8.65) into
8.18.2 Water flows at 0.32 m3/s through a 600-mm-diameter pipeline with a head loss of
0:0029 m/m. Find the value of the Hazen-Williams coeffi cient. ·
8.18.3 When water flows at 70 cfs through a 5-ft-diameter pipeline the head loss is 3.0 ft Q = (it (8.69)
per mile. Find the value of Mann'ing's n. We shall make use of these nonrigorous equations in Secs. 8.29-8.32.

8.19 NONRtGOROUS HEAD-LOSS EQUATIONS

8.20 MINOR LOSSES IN TURBULENT FLOW
If we rearrange the empirical Hazen-Williams and M anning's equations of
Sec. 8.18 into the form of head-loss equations like Eq. (8.13), they will remain Los~es due ~o l~cal disturbances of the flow in conduits such as changes in cross
explicit in all unknowns and so easy to solve, but they will of course be less ac- section, proJectmg gaskets, elbows, valves, and similar items we call minor losses.
curate than the rigorous equations that have a friction factor f that depends on In the c~se of~ very long pipe or channel, these losses are usually insignificant in
the Reynolds number, R. We can create a third, similar situation and equation if com~anson with _the loss due to pipe (wall) friction (Secs;· 5.3, 8.5) in the length
co_ns1dered. But 1f the length of pipe or channel is very short, these so-called
we use the Darcy-Weisbach Eq. (8.13) with a given, constant value off.
To get good results with these nonrigorous equations the user must select mmor losses may actually be major losses. Thus, in the case of the suction pipe of
proper values of the friction factor, either <;mv, Manning's n , or the constant f This a pump, the loss of head at entrance, especially if there is a strainer and a foot
·is more uncertain than estimating the e/D value for the Darcy-Weisbach equation valve, may be very much greater than the friction loss in the short inlet pipe.
with varyingf. In Sec. 8.18 we noted that such less-than-rigorous methods and less Whenever the average velocity of turbulent flow is altered either in direc-
accurate results are acceptable for the design of water distribution networks; they tion _or in magnitude, l~rge eddies (Secs. 4.2, 8.9) or eddy currents are set up
could also serve as first estimates for repetitive, rigorous solution procedures . causmg a loss of energy m excess of the pipe friction in that same length. 21 Head
When the friction factor is constant, i.e., it does not vary with R , we find in l~ss in decele_rating (i.e., diverging) flow is much larger than that in accelerating
all three cases that we can conveniently represent the head-loss equations in the (1.e., convergmg) flow (Sec. 8.24). In addition, head loss generally increases with
an increase in the geometric distortion of the flow. Though the causes of minor
form losse~ are usually confined to a very short length of the flow path, the effects may·
(8.65) not disappear for a considerable distance downstream. Thus an elbow in a pipe
may occupy only a small length, but the disturbance in the flow will extend for a
where we note that n is a constant exponent and not Manning's n, which we shall long distance downstream.
here represent by nm. By rearranging Eq. (8.13) when f is constant and using In the following Secs. 8.21-8.26, we describe the most common sources of
Eq. (4.7) to replace V, we obtain minor loss. There are two ways we can represent such losses. We can express
8fL them as kV2/2g, where we must determine the loss coefficien't k for each case. Or
Darcy-Weisbach K=-- n = 2 (8.66) we can represent ~hem as being equivalent to a certain length of straight pipe,
with constant f: 1r2gDs'
usually expressed m terms of the number of pipe di~meters , N. Since
By rearranging the empirical equation (8.63a) , we obtain k-
v2 f(ND) v 2
---
2g D 2g
Hazen-Williams 4.727L (8.67)
n = 1.852 k = Nf
BG units: K == CkW2D4.87 ' it follows that
_ To differentiate minor losses from other losses, we shall represent minor
For SI units we must replace the 4.727 by 10.675 in the Hazen-Williams K; recall head _losses by the symbol h' (with a prime). Because they are losses of energy,
from Sec. 8.18 the restrictions on D and V. By rearranging the empirical equa- ':e will frequen~ly rela_te_~hem to the energy line (EL) and the hydraulic grade
tion (8.64a ), we obtain hne (HGL), which we 1111tially defined and discussed in Secs. 5.11 and 5.13. Note

Manning, 4.66Ln~1 (8.68)

K=- --
D16.3 ,
n - 2 21
In lamin_a~ flow t_hese losses are insignificant, because irregularities in the flow boundary
BG units: create a mm1mal disturbance to the flow and separation is essentially nonexistent.
302 CHAPTER 8: Steady1ncompressible Flow in Pressure Conduits 8.22 Loss of Head at Submerged Discharge 303
that these losses are really only degradations of mechanical energy to a less use-

~
ful form (heat energy).

8.21 Loss OF HEAD AT ENTRANCE f.

Bell-mouthed I

Referring to Fig. 8.12, we·can see that, as fluid from the reservoir enters the pipe,
the streamlines continue to converge for a while, much as though this were a jet
issuing from a sharp-edged orifice (Sec. 11.6). As a result, we find a cross section
with maximum velocity and minimum pressure at B. This minimum flow area is
known as the vena contracta. At B, surrounding the contracted flowing stream
there is fluid in a state of turbulence but having very little forward motion. Be-
tween B and C the fluid is very disturbed because the stream expands (is less
constrained) and the velocity decreases while the pressure rises. From C to D
the flow is normal.
We see that the loss of energy at entrance occurs over the length AC, a dis-
tance of several diameters. The increased turbulence and vortex motion in this
portion of the pipe cause the friction loss to be much greater than in a corre-
sponding length where the flow is normal, as we can see from the drop of the
total-energy line. Of this total loss, a small portion h1 would be due to the normal
pipe friction (see Fig. 8.12). Hence the difference between this and the total, or
h~, is the true value of the extra loss caused at entrance.
The loss of head at entrance we can express as
(8.70)
(a) k,. = o.04
Figure 8.13
Entrance loss coefficients.
depends on how much the stream contracts as it enters the pipe. Thus it very
much depends on the conditions at the entrance to the pipe. Experiments have
determined values of the entrance loss coefficients. If the entrance to the pipe is
well rounded or bell-mouthed (Fig. 8.13a), there is no contraction of the stream
entering and the coefficient of loss is correspondingly small. For a flush or
square-edged entrance, such as shown in Fig. 8.13b, ke has a value of about 0.5.
A reentrant tube, such as that shown in Fig. 8.13c, produces a maximum con-
traction of the entering stream, because the streamlines come from around the
outside wall of the pipe, as well as more directly from the fluid in front of the en-
, trance. The degree of the contraction depends upon how far the pipe may pro-
ject within the reservoir and also upon how thick the pipe walls are, compared
with its diameter. With very thick walls, the conditions approach that of a
square-edged entrance. For these reasons, the loss coefficients for reentrant
tubes vary; for very thin tubes, ke = 0.8.

where Vis the mean velocity in the pipe, and ke is the loss coefficient, whose gen-
eral values are given in Fig. 8.13.
The entrance loss results primarily from the turbulence created by the en-
largement of the stream after it passes section B, and this enlargement in turn
EL
8.22 Loss OF HEAD AT SUBMERGED DISCHARGE 22
Discharge into Still Water
When a fluid with a velocity Vis discharged from the end of a pipe into a closed
tank or reservoir which is so large that the velocity within it is negligible, the en-
tire kinetic energy of the flow is dissipated. Thus the discharge loss is

(8.71)
\12
2g
We can confirm that this is true by writing an energy equation between (a) and
(c) in Fig. 8.14. Taking the datum plane through (a) and recognizing that the
pr essure hea<l of the fluid at (a) is y, its depth below the surface, Ha = y + O +
2
V 1/2g and He = 0 + y + 0. Therefore we obtain ·
D
v2
h;, = Ha - He = -
2g
Figure 8.12 22
Conditions at entrance. We first discussed this topic in Sec. 5.12.

•
 CHAPTER 8: Steady Incompressible Flow i-n Pressure Conduits
8.23 Loss Due to Contraction 305
304
For a general streamline such as ef in Fig. 8.15, from energy Eq. (5.28) we have
2 2
Pe
-+z .
+ V- ] -h 1 = [Pt -+z +-Yc ]
7\,____..__::.r.-.....--
I • .
[ Y e 2g d y f 2g

Substituting from Eq. (8.72) into (8.73) and rearranging we obtain

(8.73)

I
_y
l 1_
y2 Yc2
--- (8.74)
:t - (b)
.,
I ... - :: :: :: ::
2g · 2g
-+-
__:.....:;.----
_
(!12=====-=
. -----
Figure 8.14
Loss due to submerged This equation states that the discharge loss is equal to the difference between the
discharge into still water. discharge and ultimate velocity heads. Equation (8.71) is a special case of this
equation.
. . . k = 1 0 ·under all conditions; so the only wa~
Thus the discha~ge loss co~ffi~1en\educe. the magnitude of V by means of a di-
to reduce the discharge loss is to d' . . g draft tube that discharges the
verging tube. This is the reason for a ivergm EXERCISES
. flow from a reaction t~rbine (Sec. i6.8).note that discharge loss occurs after the 8.22.1 A 12-in-diameter pipe (f = 0.02) carries fluid at 8 fps b~tween two tanks. The
As contrast~d ~jth he_nltran~ean::,loss occurs· after the fluid enters the pipe. entrance and exit conditions to and from the pipe are square-eQged and flush
fluid leaves the pipe, w 1 e en r with the wall of the tank. Find the ratio of the minor losses divided by the pipe
friction loss·if the length of the pipe is (a) 4 ft; (b) 80 ft; (c) 1600 ft.
Discharge into Moving Water 8.22.2 A 375-mm-diameter pipe (f = 0.017) carries fluid at 3.6 mis between two tanks.
Let us no\: find th~ head loss lw
water movmg away ma channe 'sue
.
1e:
1 ·. a submerged discharge enters a body of
as from a culvert with a submerged outlet
. b. (S 16 8)
The entrance and exit conditions to and from the pipe are reentrant. Find the
ratio of the minor losses divided by the pipe friction- loss if the length of the pipe
·1 f react10n tur me ec. · · is (a) 2 m; (b) 50 m; (c) 1000 m.
(Sec. 10.23), or as_in the tai ra~_e o a t aight and parallel, so, as demonstrated
At the outlet, the str~~°: m~.s ar:c!iss it must be hydrostatic. Afte~ a suf~- 8.22.3 Water leaves a turbine at 18.5 fps and enters a tailrace having an average
in Sec. 5.A, the pressure disrnbut10n t· f i·n Fi·g 815 the streamlmes will velocity of 1.5 fps. (a) What is the submerged discharge head loss? (b) By what
..
cient _distance :om
f th utlet as at sec ion c
_e,_o . ' , ·u 1 d the flow ·velocity
· , .
will be practlc~lly
percentage is this loss reduced if tpe provision of a draft tube increases the
discharge flow area to six times the size? ·
again b_e essenti_all~ straight and pa~~ e ~:nso the pressure distribution at section
uniform(=:= -i-;;) over the f~~} c~~n;:reftre ~er Eq. (3.6) (using gage pressures) 8.22.4 A smooth 300-mm-diametei pipe is 90 m long and,has a flush entrance and a
submerged discharge. It carries 15°C water at a v~locity of 3 mis. What is the
cf must also be hydrosta~1e,_an .
total head loss?
Pe + Ze = Pt + Zf = Pc + Zc = 0 (8.72)
y y y 8.22.5 Oil with a kinematic viscosity of 0.001 ft 2/sec and a specific gravity of 0.92 flows
at 8 fps through a smooth 15-in-diameter pipe which is 400 ft long with a flush
entrance and submerged discharge. What is the head loss in feet of oil and in psi?
_ _ _ _. - - . be a major factor, greater accuracy is
23 In a short pipe, where the ~1scharge loss may lained in Sec. 5.1 (see also Eq. 8.45a).
obtained by using the correction factor a, as exp
8.23 Loss DUE TO CONTRACTION
__:;;:__....--t Datum (c

'::> . i
,Sudden Contraction
I
The phenomena accompanying the sudden contraction of a flow are shown in
) _. ~
Fig. 8.16. There is a marked drop in pressure due to the increase in velocity and
to the loss of energy in turbulence. Note that in the corner upstream at section
C there is a rise in pressure because the streamlines here are curving, so that the
centrifugal action causes a greater pressure at the pipe wall than in the center of
the stream. The dashed line indicates the pressure variation along the central
Figure 8.15 . · t streamline from sections B to C.
Loss due to submerged discharge mto movmg wa er.
306 CHAPTER 8: . 'bl e Flow in Pressure Conduits
Steady fncomp, .essr
8.24 Loss Due to Expansion 307
The nozzle at the end of a pipeline (see Fig. S8.l lb) is a special case of
gradual contraction. An equation like Eq. (8.75) also governs the head loss
through a nozzle at the end of a pipeline, where k, becomes the nozzle loss co-
kn whose value commonly ranges from 0.04 to 0.20 and V, = i-; is the jet
efficient24
-- El
velocity. However, we can not regard the head loss through a nozzle hn as a
minor loss, because the jet velocity head is usually quite large (see, e.g., Sample
Prob. 8.12). More details on the flow through nozzles is presented in Sec. 11.6.

8.24 Loss DUE TO EXPANSION

Sudden Expansion

The conditions at a sudden expansion are shown in Fig. 8.17. There is a rise in
pressure because of the decrease in velocity, but this rise would be even greater
Vz--+-- if there were not the loss in energy. There is excessive turbulence in the flow
F from C to F, beyond which the flow is normal. The drop in pressure just beyond
section C, wqich was measured by a piezometer not shown in the figure, is due
to the fact that the pressures at the wall of the pipe are in this case less th an those
Figure 8.16 . d l ) in the center of the-pjpe because of centrifugal effects.
Loss due to sudden contract10n. (Platte to sea e.
Figures 8.16 and 8.17 are both drawn to scale from test measurements for
the same diameter ratios and the same velocities, and they show that the loss
TABLE 8.2 Loss coefficients for sudden contraction due to sudden expansion is greater than the loss due to a corresponding con-
traction. This is so because of the inherent instability of flow in an expansion,
0.0 0.7 0.8 0.9 1.0 where diverging flow paths encourage the formation of eddies within the flow.
0.50 0.45 0.42 0.39 0.36 0.33 0.28 0.22 0.15 0.06 0.00 Moreover, separation of the flow from the wall of the conduit creates pockets of

24
See also Eq. (11.14).
. C ..
E the conditions . s1m1
are . ·1 ar to those described for
. entrance
b
8.21). Wetocan, represent t 11e 1oss o f head for a sudden contraction Y
(Sec. From

h'C (8.75)

k h th values given in Table 8.2. / - 0

whereThe as e 1oss o f S e c · 8 ·21 is a special case where D2 D1 - .
c entrance

G dual Contraction
--l_
l ___ ---~-~
wall
--™™""',1----.,i _____
ra . we should avoid abrupt changes of cross ~ ·~ ., ~ ::~_: .?~~'..-,---
In order to reduce the foreg~mg lo~es, ·n from one diameter to the other with
Vi--+-
section. We could achiev~ _this by c_ anthe ~rustrum of a cone. With a smoothly
a smoothly curved transition ~r with 11 0 05 is possible. For conical re-
..
curved trans1t1on, a 1oss c oeffic1enO.
t kc as sma
"ble as . 20-400 .
• · k f about O 1 1s possi , with a total cone angle of Figure 8.17
F
larger totaclocone ang ies result in higher values of kc
ducers, aorm1111mum
Smaller
Loss due to sudden enJargemen t. ( Plotted to scale. Velocity is the same as in F,g. 8.16.)
308 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
8.24 Loss Due to Expansion 309
We usually assume
den enlargement is that p ' -- P1, m
· wh ich
· case the loss of head due to sud-

h' = (½ - ½)2 (8.76)

X 2g

Although it is possible that under some d. . , .

Figure 8.18 also possible for it to be either mdre or less thanct~n t itI~ns, _P w1_ll equal Pi, it is
of head will be either less or more tha th t . ~ va ue, m which case the loss
of p' will depend on the manner in nh_ : tt1v~n. y Eq._ (8.76). The exact value
corner adjacent to this annular rin ~ ic e md ed~1e~ move around in the
eddying turbulence outside the flow region. In converging flow, on the other quite small and of negligible impo/ta. owever, the deviation from Eq. (8.76) is
hand, there is a dampening effect on eddy formation, and the conversion from Th d. nee.
pressure energy to kinetic energy is more efficient. e ischarge loss into still water of S 8 22 . .
infinite compared with A or V: _ h ec. · is a special case where A 2 is
We can derive an expression for the loss of head in a sudden enlargement 0
On occasion we mar need t~ e~ s~: at ~q. \8.76) will re_duce to Eq. (8.71).
as follows. In Fig. 8.18, section 2 corresponds to section Fin Fig. 8.17, which is a locity (see, e.g., Sec. 8 27) If p ss this m1?-or_ loss all m terms of one ve-
section where the velocity profile has become normal again, and marks the end to obtain . . so, we can use contmmty, (n/4)Df½ = (n/4)Df½,
of the region of excess energy loss due to the turbulence created by the sudden
enlargement. In Fig. 8.18 let us assume that the pressure at section 2 for ideal
flow (without friction) is p 0• Then from Eq. (5.30) in this ideal case (½ - ½)2 = (1 - Df)2v;2
Di i
= (DI
Df - 1)2u2
v2
(8.77)

Po
=P1- +V;
- -Vi
- Gradual Expansion
y y 2g 2g
To minimize the loss accompanyino a reduction in v 1 .
such as that shown in Fig 8 19 Th bd.ff e oc1ty, we can use a diffuser
If in the actual case, however, the pressure at section 2 is Pi, while the average b · · · e 1 usermayhave d ·
ea ~rustum of a cone (straight sided). In Fi 8 19 ha curve outlme, or it may
pressure on the annular ring is p' then, equating the resultant force on the body function of the angle of diver e d g. . t e loss of head will be some
of fluid betwe~n sections 1 and 2 to the time rate of change of momentum be- length of the diffuser being defe;~1n:~ byal:ho of tthe rati? of t?e two areas, the
tween sections 1 and 2 per Eq. (6.7a), we obtain I fl th ese wo vanables
n ow rough a diffuser we can c ·d h ·
components. One is the ordinary . pipe-fi~~~o~\~s:
, , twotha_ll
hoss
ic we canas made up of two
represent by
f v2
From this,
ht=
I --dL
D 2g
In order to integrate
as functions of L Fortheo fore goin g, we need t_o ~xpress the variables f, D' and V
P2 A1 P1 A2 - A1 p' A1 V; Vi
=- -+----+---- . ur present purpose, it is sufficient, however, merely to
y A2 Y A2 Y A2 g g

The loss of head is caused by friction, and so it is given by the difference between
the ideal and actual pressure heads ( or total heads) at section 2. Thus h_~ =
CPo - P2VY, and noting that
A1V1 = A2V2
and that A Vi2 = A 1V1Vi = A 2V2Vi, we obtain, from substituting the preceding
1
expressions for p 0 /y and p 2 /y into (p0 - P2)/y,
Figure 8.19
Loss due to gradual enlargement.
 310 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
8.24 Loss Due to Expansion 311
0.3
\
\
\
\
' -

' ._..,,
/
/
I
I
I
I
1.2 ~ - - - - - --

1.0
- --~

I
J.
SAMPLE PRonm_1 8.10·- -~-~e:~::~~r:;~a~:S~-=~-d::::=
t? 200 mm diameter, what is tlie ~ead
100 fls. (a) If the pipe suddenly e~la~ges
0.2 0.8
r lohss. (b) If the same enlargement is via a diffuser with a total cone anole of 200
~
i w at 1s the head loss then? b ,

0 0 ~
<J)
~ 0.6 I Solution
Cl)
:J :J
ai co r Q 4Q 4(0.100)
> > (a) l7i =
0.1 0.4
A1
= = 5.66 mis
nDl n(0.15) 2
0.2 4(0.100)
½ n(0. 2 0) 2 = 3.18 mis
0 ""-------'------'--~-------' 0 ~--'----'----L-----'----'------'
0 4 8 12 0 20 40 60 90 120 180 (5.66 - 3.18) 2
Cone angle a Cone angle a h'X = = 0.312 m ANS
2(9.81)
(a) (b)
( b) Fig. 8.20 for a = 20°: k' = 0.38
Figure 8.20
Loss coefficient for conical diffusers. Eq. (8.78): h' = 0.38(0.312) = '0.1187 m ANS

note that the friction loss increases with the length of the cone. Hence, for given
values of D 1 and D2 , the larger the angle of the cone, the less its length and the EXERCISES
less the pipe friction, which we can see from the curve marked Fin Fig. 8.20a.
However, in flow through a diffuser there is an additional loss component, due 8.24.1 T\~O pipes w~th a diamete~ ratio of 1:2 are connected in series (Fig. X8.24.l).

, to turbulence set up by induced currents that produce a vortex motion over and
above that which normally exists. This additional turbulence loss will naturally
increase with the degree of divergence, as we can see from the curve marked T
With a velocity of 6:8 mis m the smaller pipe, find the Joss of head due to
(~) sudde1: contract10n; (b) sudden enlargement; (c) expansion in a conical
diffuser with a total angle of 30°, and of 10°.
in Fig. 8.20a, and if the rate of divergence is great enough then the flow may sep-
2D
arate from the walls with eddies flowing backward along the walls (see Sec. 9.6). -7.._£__ D 2D D . 2D
~ :::_------
The total loss in the diverging cone consists of the sum of these two losses, -... _.,..
----,__
------
....... 30°, 10°
marked k'. This has a minimum value at 6° for the particular case chosen, which __r--
is for a very smooth surface. If the surface were rougher, the value of the friction Figure XS.24.1 (a) (b) (c)
Fwould be larger. This increases the value of k', as in the dashed curve, and also
shifts the angle for minimum loss to 8°. Thus the best angle of divergence in- 8 24
- -2 A 5-in-diameter pip~(/= _0.033) 110 ft long connects two reservoirs whose
creases with the roughness of the surface. water-sur~ace elev~tions differ by 12 ft (Fig. X8.24.2). The pipe entrance is flush
We have seen that the loss due to a sudden enlargement is very nearly a~1d the discharge is s~bmerge_d. (a)_ Compute the flow rate. (b) If the last 10 ft of
equal to (i~ - ½)2/2g. The loss due to a gradual enlargement is pipe were replaced with a corneal diffuser with a cone angle of 10° compute th
flow rate. ' e

(½ - Yi)2
h' k'---- (8.78)
2g

Values of k' as a function of the cone angle a are given in Fig. 8.20b, for a wider
range than appears in Fig. 8.20a. Note, interestingly, that at an angle slightly ..._____(-'---h-'-)_D_=_S_in__:_,_
l _.-_- _..10
:._.:.:0:..:ft:..:__ _~
1
above 40° the loss is the same as that for a sudden enlargement, which is 180°, and 100
that between these two the loss is greater than for a sudden enlargement, being a
maximum at about 60°. This is a result of the induced currents that develop.
Figure XS.24.2
312 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
8.26 Loss in Bends and Elbows 313
8.25 Loss IN PIPE FITTINGS A

For pipe fittings we can express the loss of head as kV 2/2g , where Vis the veloc-
ity in a pipe of the nominal size of the fitting. Table 8.3 gives typical values of k.
As an alternative, we may account for the head loss due to a fitting by increasing
the pipe length by amounts given by L/D in the table. Howeve_r, we should rec-
ognize that these fittings create so much turbulence that the loss caused by them
is proportional to V 2 , and so we should restrict this latter method to cases where
the pipe friction itself is in the zone of complete turbulence (fully-rough-pipe
flow). For very smooth pipes it is better to use the k values when determining
the loss through fittings.

TABLE 8.3 Values of loss factors for pipe fittings

Fitting k L/D
Globe valve, wide open 10 350
Angle valve, wide open 5. 175
Close-return bend 2.2 75
T, through side outlet 1.8 67 Inside of bend
Short-radius elbow 0.9 32 I

Medium-radius elbow 0.75 27 B

Figure 8.21
Long-radius elbow 0.60 20 Secondary flow in a pipe bend.
45° elbow 0.42 15
Gate valve, wide open 0.19 7
half open 2.06 72 downstream from the bend. In fact more th . .
elbow occurs in the straight pipe following itn half the fnct10n loss of a bend or
The pressure difference that dev 1 . b
the outside and inside of the bend . e op~ etween A and B (Fig. 8.21), on
(Sec. 11.16), which measures dischar~~~ use as the basis of the elbow meter
8.26 Loss IN BENDS AND ELBOWS We may eliminate most of the head loss due
vaned elbow, such as that in Fig 8 22 Th . to a sharp ?end by using a
In flow around a bend or elbow, because of centrifugal effects (Sec. 5.16), there secondary -flows that would oth . : .d e vanes impede the formation of the
N erw1se evelop
is an increase in pressure along the outer wall and a decrease in pressure along ote that the amount of h d I h .
the 'inner wall. In Fig. 8.21 we see the centrifugal forces acting on a number of given by ea oss t at a pipe bend or elbow causes, as
fluid particles, each of mass m on pipe diameter CE that is normal to the pipe's
plane of curvature. The centrifugal force m V 2/r on the partic;ks near the center v2
hb = kb-
of the pipe, where the velocities are high, is larger than the centrifugal force on 2g (8.79)
the particles near the walls of the pipe, where the velocities are low (see
Fig. 8.10). Because of this unbalanced condition, a secondary flow 25 develops, as
shown in the cross section of Fig. 8.21. This combines with the axial velocity to ((----
form a double spiral flow, which persists for some distance. Thus not only is (, Turning vanes
there some loss of energy within the bend itself, but this distorted flow persists
for some distance downstream until viscous friction dissipates it. The velocity in fr) ,.
rr
the pipe may not become normal again within as much as 100 pipe diameters

Figure 8.22
25
Secondary flow in the bends of open channels is discussed in Sec. 10.21. Vaned elbow.
 8: Steady Incompressible Flow in Pressure Conduits
314 CHAPTER
1.0 . . - - - - - - - - - - - - - - - - ~
8.27
0.8
0 .6
0.4
1.5 2 3 4 5 6 7 8 910
r/D
Figure 8.23 f · /2 · th b d
Resistance coefficients for 90° bends (resista~ce due to .length o pipe 1rr m e en
must be added, where r is the radius to the pipe centerlme ). (From Ref 46.)
is the amount in excess of the loss for an equal len~th of straight pipe. It depends
strongly on the ratio of the radius of curvature r (F.1g. 8.21) ~o the diameter of the
· D 'Note also that we cannot treat combinations of different bends placed
~:r:e t~gether by 'adding up the losses of each one considered separately. The
total loss depends not only on the spacing between the bends, but also o~ the
relative directions of the bends and the planes containing them. Bend loss 1s not
proportional to the angle of the bend; for 22.5° a nd 45° b~nds the lo~ses are re-
0
spectively about 40% and 80% of the loss ~n a 90 b_end. Figure 8.23 gives values
of kb for 90° bends, varying with bend radms and pipe roughness.
EXERCISES
8.26.1 Water at 72°F flows through a 110-ft-long, 5-in-diameter wrou~ht-iron_ pipe that
contains the following fittings : one open glob_e val~e, one mednm:-:adms elbow,
and one 90° pipe bend (kb= 0.13) with a radms of curvature of 4) 1~. Th~ length
of the bend is not included in the 110 ft. There ?re_ no entrance or d1scha1 ge
losses. Find the total head loss if the flow velocity 1s 4.8 fps.
8 _26 .2 Water at 18oC flows through a 25-m-long, 75-mm-diameter commercial steel_
pipe that contains the following fittings: one open angle valve, one short-radms
elbow, and one 90° pipe bend (kb = 0.12) with a radius of curvature of 600 mm.
The length of the bend is not included in the 25 m. There.ar~ no entrance or
discharge losses. Find the total head loss if the flow velocity is 2 mis.
8.27 Single-Pipe Flow with Minor Losses 315
SINGLE-PIPE FLOW WITH MINOR LOSSES
We have examined the fundamental fluid mechanics associated with the fric-
tional loss of energy in single-pipe flow, caused by both the wall roughness of the
pipes and by pipe fittings and the like that disturb the flow (minor losses). While
the interest of the scientist extends very little beyond this, it is the task of the en-
gineer to apply these fundamentals to various types of practical problems.
Specialists generally concede that for pipes longer than 1000 diameters, the
error from neglecting minor losses is less than that inherent in selecting a value
for the friction factor (f, n, or CHw). In applying this rule, one must of course use
common sense and recall that a valve, for example, is a minor loss only when it
is wide open; partially closed, it may be the most important loss in the system.
When minor losses are negligible, as they often are, we can solve pipe-flow prob-
lems by the methods of Secs. 8.13-8.19.
When we include minor losses, the total head hL loss between two points is
the sum of the pipe (wall) friction head loss h1 plus the mipor losses, or
(8.80)
We can represent the pipe friction loss by a number of different equations, and
it may depend on a number of different factors, as noted in Secs. 8.1-8.19. And,
as we noted in Sec. 8.20, we can represent the minor losses as a coefficient mul-
tiplied by the velocity head, kV2/2g, or as an equivalent length of pipe, expressed
as a number of diameters, ND. As a result, the right-hand side of Eq. (8.80) can
take on many forms. Because of the extra term added by the minor losses, such
forms of this equation are often more challenging to solve than those that omit
minor losses (Secs. 8.14-8.19).
Nonrigorous Equations. In problems where a constant f value is given,
Eq. (8.80) still has only one unknown, namely, hL or V or Q or D. In most cases
this equation is explicit in the unknown (Appendix B), and so is easy to solve.
But for sizing problems, the resulting equation in Dis of the fifth degree and im-
plicit, requiring trial and error or an equation solver as discussed in Sec. 8.17.
The empirical equations (Secs. 8.18-8.1.9) do not necessarily continue to
yield direct solutions when minor losses are included. By including kV2/2g or
ND in Eq. (8.65) with Eqs. (8.67) and (8.68), we find that these equations are al-
ways explicit for Type 1 (find hL now, not just h1), usually explicit for Type 2 (find
Q or V), and never explicit for Type 3 (find D). The exception with Type 2 oc-
curs when using the Hazen-Williams equation and representing minor losses by
kV2/2g; this is because, in Eq. (8.65) with (8.67), we find that the pipe friction
loss is proportional to vi. 852 , so the two terms contain V to different inconve-
nient powers. Type 3 problems are more complex in part owing to the fact that
Q is given rather than V, since this causes V to also be a function of D. We first
discussed these three problem categories in Sec. 8.15.
Rigorous Equations. In practice, we may need to take into account the varia-
tion off, which can vary by as much as a factor of five for smoother pipes
316 CHAPTER 8: Steady Incomp,:essible Flow in Pressure Conduits 8.27 Single-Pipe Flow with Minor Losses 317
(see Fig. 8.11). When including minor losses, methods involving the Darcy-
Weisbach equation (8.13) and a·n unknown friction factor f remain explicit or
direct for Type 1 problems (find hL), but for the other two types of problems
~~r ~~~;) !~;%tI)~i !rt~~~~t;:;~~1;; : ~~:rs~~dt~~~~::ut~~~~~!s
ur u ent. If R < 2000, the flow is laminar (S 8 2) d . ow is
~o u~e
(find Q or V or D) they are always implicit (Appendix B). The other governing the problem wit~ Eq. (8.28) and Eq. (4 _3) if n:~d~d.' an we must mstead solve
equations are the same as in Sec. 8.14. th T:e f~llol~mg sam_ple pro?km illustrates the method of solution for flow
roug a pipe me of umform diameter with minor losses.
Implicit Equations." To solve the many implicit equations that arise when we
include minor losses, it is clear we must use iterative procedures, either manu-
ally or programmed. Manual iterative procedures (trial and error) for this pur-
pose are generally like those described in Sec. 8.15. But occasionally it will be
easier to make trials for D rather than f SAMPLE PROBLEM 8.11 Water at 60°F flows h · ·
Automated iterative procedures are available in equation solving software cast-iron pipe sketched in Fi S811 Th . t !ough the new ~0-m-diameter
packages~ and in programmable calcula~ors with an equation solving function, as
discussed in Appendixes C and D and Sec. 8.17. These unquestionably are the
sharp-~ornered but nonproje~ting · an~ Liz ~p~~i ;;
only Fig. 8.11 and a basic scientific ca
i~o~
f~ long, its entr~nce is
. . m t e fl~w rate usmg (a)
most convenient, and we can expect them to become more accessible in time. programmable calculator (c) equat1·0 lclu!ator,f(b) an equat10n solver on a
' . n s~ vmg so tware such as Mathcad.
When using an equation solving software package like Mathcad or Excel
(Appendixes C and D), we just need to specify n equations in n unknowns, as
described in Sec. 8.17. We merely hav·e to type in with the energy equation the
terms added to account for minor losses, and then assign values to any new loss
coefficients. Otherwise, the procedure is unchanged. -
To use an equation solver in a programmable calculator, we recall from
Appendix C.1 and Sec. ~.17 that for the implicit, nonlinear equations we must
deal with there can be only one equation in one unknown. Similarly to the de-
-1~~~~=-==-~~~::::=!~t-=lT
velopment of Eq. (8.56), we can develop a "universal" turbulent flow equation
for use in an equation solver, including minor losses, by eliminating h1 andffrom ~: li.\z
Eq. (8.80) with the help of Eqs. (8.13), (8.51), and (4.7). Expressing minor losses
2;h' in terms of 2;kV2/2g,_we obtain 2
----.ti----;l.
3
(a) Free discharge
Turb1,1lent flow: ( b) With nozzle
Figure S8.11
Discharge from a reservoir. As L/D gets larger the EL and HGL h
-2 -4L/D
- - - = -2log (e/D
- + 2.5lnvD (8.81) one another. approac
n gD hL 3.7 4Q
- - - - "2:,k
8Q2 Solution
This is a Type 2 problem with minor losses.
If we express minor losses in terms of an equivalent length of pipe, ND, then we
obtain Table 8."1 for cast iron: e = 0.00085 ft; e/D = 0.00085/(10/12) = 0.00102
Turbulent flow: Sec. 8.21 for the square-edged entrance: ke = 0.5
4Q _(L_/D_)_+_N_ = _ 2 log(-e/D_ + _2._51_v /(L/D) + N) (8.82)
Energy Eq. (5.28) from the water surface to the free jet at point 2:
nD 2 2ghL 3.7 D \f 2ghL
Each of these equations involves eight variables: hv Q, D, L, e, v, g, and 2;k or 260 + o+ o- (o.5 + f 5000 ) ½2 = o + o + ½2
(1)
10/12 2g 2g
N. We note that the equations are implicit in hv Q, and D, but, with an equation
solver we may easily solve for any one of these,three quantities if the rest of the
variables are known. Similar and less complex "universal" equations can also be
developed using the Haaland equation (8.52); however, they sometimes yield
false results when ,large diameters cause the argument of the logarithm to be
== th~o~rammed computing ajds (Appendix C) could help solve problems marked
Is Icon. ,
 8: Steady Incompressible Flow in Pressure Conduits
8.27 Single-Pipe Flow with Minor Losses 319
318 CHAPTER

2(32.2)260 (c) Using equation solving software such as Mathcad (Appendixes C.2-C.3 and
260
(a) Rearranging: = 1.5 + 6000.f' D.2- D.3); we es tablish ~ "so!ve block" containing four equations as in Sample
1.5 + 6000f Prob. _8.9b , except that m this case we rep!ace Eq. (8.13) by preceding energy
equatl~n (1). Also, we must cha~ge the "Fmd" statement to: Find (f, V, R, Q).
Fig. 8.11, right-hand side, for e/D = 0.00102: fmin = 0.0196 We_assign_ the seven know~ vanable~ the same values as in solution (b ), and
assign estimated values (without umts) to the four unknown variables, say
Try f = 0.0196: V2 = V 16,744/ [l.5 + 6000(0.0196)] 11.86 fps f = 0.030, V = 5 fps, R = 100,000, and Q = 10 cfs.
The program delivers the result:
D"V = 10(11.86) = 11 8.6
ij
f = 0.0201, V = 11.71 fps, R = 801,731, Q = 6.39 cfs ANS

i
Fig. 8.11 for e/D = 0.00102 and D"V = 11_8.6: f = 0_.020. Obtain_ed ~nd assumed
f values are different, so we must try agam. Tabulatmg all the tnals. . I Because this R is for turbulent flow, the equation (8.51) we used for f is valid
and so therefore is the answer we obtained for Q.
Note the simplicity and clarity of the Mathcad solution (c) in comparison to
'
the other two solutions.
Try f V , ft/s D"V Obtained/

11 .86 118.6 0.020 Try aga in

0.0196
11.74 117.4 0.020 Converged! If the pipe in this example discharged into a fluid that was at a pressure
0.020
other than atmospheric, we would have to use the proper value of p 2/y in the
energy equation.
2
Q = -n (10)
- (11.74) = 6.40 cfs ANS Another example of flow from a reservoir is that of a pipeline (known as a
Eq. (4.3): 4 12 penstock) leading to an impulse turbine (Sec. 16.2). In this case the pipe does not
discharge freely, but ends in a nozzle (Fig. S8.llb ), which has a known or as-
Alternatively, replace use of the chart by equations as follows: sumed loss coefficient, kn. The head loss in the nozzle, hn, is associated with the
high issuing velocity head, ½'2/2g, and is therefore not a minor loss (Sec. 8.23).
Table A.1 for water at 60°F: v = 1.217 x 10- 5 ft 2 /sec
We use the equation of continuity to write all losses in terms of the velocity head
in the pipe. This is the logical choice for the "common unknown," because we
Eq. (8.54) for e/D = 0.00102: fmin = 0.01973 will again build the trial-and-error solution around the pipe friction loss rather
Energy equation with assumed f gives V {as before); R = DV/v than the nozzle loss.
Insert values of e/D and R into Haaland equation (8.52) to get a calculated f;
compare with assumed f. If the difference is less than about 0.5%, we have
convergence; otherwise repeat.
(b) Using an equation solver on a pro 0 rammable calculator (Appendixes C.1 and
• r~;::::-;::::;;-~~i~~t~h:;p~~n::
Sa~ple _Prob. 8.~l 1s now fitted with a nozzl~, at the end of which discharges a
D.1), follow the procedur~ of Sampl~ Prob. 8.9a, which_ eliminates f and Rand V, I 2.~-m-diameter Jet and that has a loss coefficient of kn = 0.11. Find the flow rate
but with an equation that includes mmor losses. Equation (~.81) do~s not appear
to be appropriate, because it does not include for the chang~ m velocity head from
zero at point 1 to ½2/(2g) at point 2. However, from prec~dmg Eq. (1) we see that
II
t
usmg only Fig. 8.11 and a basic scientific calculator.
Solution
Referring to Sample Prob. 8.11, let point 2 now be on the pipe at the base of
we can include for this by increasing ~ k by one. Assign to the , seven 2
known
variables the values (without units) g = 32.2 ft/sec2, v = 1.217 x 10-s ft /sec, hL =
! the
i
nozzle and_ J?Oint 3 be in the jet. The head loss in the nozzle is 0.11 Vl/2g
(Sec. 8.23). Wntmg the energy equation (5.28) between 1 and 3,
L1z = 260 ft, °'Z k = ke + 1 = 1.5, L = 5000 ft, e = 0.000 85 ft, and D = 10/12 ft. ~-
Assign an estimated value of 10 cfs (say) to the unknown Q. .
Instruct the solver to solve Eq. (8.81) for Q. The result 1s Q = 6.39 cfs.
r
; 260 + 0 + 0 -
( v22 v22
0.5 - + 6000f- + 0.11-
¼2)
3
= 0 + 0 + -3
v2
2 ~ 2g 2g 2g 2g
But we must check the R. Equation (4.7): V = 4Q/(nD ) = 11.71 fps, so R =
DV/v = 8.02 x 105 , and the turbulent flow assumption is correct. Therefore

Q 6.39 cfs ANS

fffl: Programmed computing aids (Appendix C) could help solve problems marked
with this icon.
320 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits

r---~.~-_
/-·--: ·-
8.28 Pipeline with Pump or Turbine 321
Using continuity Eq. (4.17), ½/2g2
= (10/2.5)4v;2/2g =
•
v2
2
256½ /2g. Thus
V ~ k r
260 = (0.5 + 1.11

Select a trial value off. Let f =

X 256 + 6000!) ~
2

0.02 for the first assumption. Then 260 = (285 +

- · · Figure XS.27.1
.-- ~
--~=~1------B~-0
' . ,, I
~
120)Vi/2g, from which 8·27·2 A 180-mm-diameter pipeline (/ = o
2 0 .
260 diameter water jet into the at h. 32) 150 m long discharges a 60-mm-
½ = = 0.643 ft surface at intake (Fig X8 27 lm) oTsph ere at a point that is 80 m below the water
2g 405 · · · · e entrance t th · ·
ke = 0.9, and the nozzle.loss coefficient is 0 055 oF" e pipe IS reentrant, with
and ½ = \!(2(32.2)0.643 = 6.43 fps. With D"V = 10 x 6.43 = 64.3 and e/D = pressure head at the base of ( t ) . . md the flow rate and the
en ranee to the nozzle.
0.001, Fig. 8.11 shows f = 0.02. In this case we can consider the first solution
8.27.3 A horizontal 4-in-diameter pipe (f = 0 028) . .
sufficiently accurate, but in general the value off we determine from the chart below the surface (Fig X8 27 3) C : •. . projects mto a body of water 2.5 ft
may be quite different from that we assume, and we may need to make one or · - · · · . ons1dermg all Ios
pomt 1.) ft from the end of the i .f I
fi d h
. . ses, n t e pressure at a
two more trials. (a) into the body of water (b) i/te f1 tht1ebvedloc1ty is 12 fps and the flow is
The rate of discharge is Q = A 2½ = 0.545 x 6.43 = 3.51 cfs ANS ' 0 e o y of water.

Then ½= 16½ = 16 X 6.43 = 102.9 fps

I
\7
In addition, H 2 = pJy + Vi/2g = 260 - (0.5 + 0.02 x 6000)0.643 = 182.58 ft, and y
t
___f V
the pressure head pJy = 182.58 - 0.643 = 181.93 ft. • .1
This example shows that the addition of the nozzle reduces the discharge D.f
from 6.40 to 3.51 cfs, but increases the jet velocity from 11.74 to 102.9 fps. The head
Figure XS.27.3
1
loss due to pipe friction is 77.1 ft and the head loss through the nozzle is 18.09 ft.
8.27.4 A horizontal 100-mm-diameter i e ( = 0 . .
below the surface (Fig X8 27 3)PCP fd .. 027) projects mto a body of water 1 m
• · · · • ons1 ermg all Ios fi d h
pomt 5 m from the end of the pipe if the veloc· . ses, n t e press~re at a
the body of water (b) out of th b d f ity is 4 m/s and the flow is (a) into
' e o yo water.
When solving Type 3 sizing problems, .in general, the diameter we obtain 8.27.5 A 450-ft-long pipeline runs between tw .
will not be a standard pipe size, and the size we select will usually be the next
largest commercially available size. In planning for the future, we must remem- th:
water, and the intake end is square-ed ~dr:~e;:01rs, b~th ~nds being under
between the water surface levels of t onp_roJ~ctmg. The difference
t~e discharge if the pipe diameter is 12 . wo ~ese~voirs is 150 ft. (a) What is
ber that scale deposits will increase the roughness and reduce the cross-sectional
area. For pipes in water service, the absolute roughness e of old pipes (20 years pipe is old, assume that the growth oft I~ an 1 f;
0.028? (b) When this same
11.25 in and that f = 0 06 Wh t ·11 h u ere es 1_as reduced the diameter to
and more) may increase over that of new pipes by three-fold for concrete or ce- . . a WI t e rate of discharge be then?
ment-lined steel, up to 20-fold for cast iron, and even to 40-fold for tuberculated 8.27.6 Solve Exer. 8.27.5a when f is unknown but aive - .
wrought-iron and steel pipe. Substituting V = 4Q/(nD 2) into Eq. (8.13) shows water temperature is 60oF Wh t h
·
. b n that e - 0.005 ft and that the
a , t en, IS the value off?
that for a constant value off, Q varies as D 512 • Hence for the case where minor
losses are negligible and f is constant, to achieve a 100% increase in flow, we
need to increase the diameter by only 32%. This amounts to a 74% increase in
cross-sectional area.
8.28
PIPELINE WITH PUMP OR TURBINE

!fa pump l~fts _a fluid from one reservoir to anoth . .

lt _d~work m lifting the fluid the heioht L'.1 b . er, as m Fig. 8.24, not only does .
EXERCISES fnct10n head loss :f,h (includ· b_ z, ut it ~lso has to overcome the total
· r . L mg mmor losses) m the
pipe mes. This head loss is equivale t t
t"
. sue wn and discharge
8.27.1 An 8-in-diameter pipeline (f = 0.028) 500 ft long discharges a 3-in-diameter
water jet into the atmosphere at a point that is 250 ft below the water surface at n o some added hft, so that the effect is the
intake (Fig. XS.27.1). The entrance to the pipe is reentrant, with ke = 0.9, and
the nozzle loss coefficient is 0.045. Find the flow rate and the pressure head at
the base of (entrance to) the nozzle. lffiiJ: Programmed computing aid (A .
with this icon. s ppendix C) could help solve problems marked
322 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits 8.28 Pipeline with Pump or Turbine 323
F Vf/2g, where ½ is the velocity of the jet. Thus the total pumping head is now
E
(8.84)

In any case we can determine the total pumping head by writing the energy
equation between any point upstream from the · pump and any other point
downstream, as in Eq. (5.37). For example, if the upstream reservoir were at a
higher elevation than the downstream one, then the L1zs in the two foregoing
equations would have negative signs.
B The machine we use to convert flow energy into mechanical work we call a
turbine. In flowing from the upper tank in Fig. 8.26 to the lower, the fluid loses
potential energy head equivalent to L1z. Part of this energy is lost to hydraulic
F~ure8~ .
Pipeline with pump between two reservoirs. friction in the pipe, and the remainder reaches the turbine. Of that part which
reaches the turbine, part is lost in hydraulic friction within the turbine, and the
rest converts into mechanical work.
The power that reaches the turbine is the initial power minus the friction
same as if the pump lifted the fluid a height L1z + "2:hL. Hen~e the pow:r losses (pipe plus minor) in the pipeline, or yQ(& - 2,hL). The power the tur-
the pump delivers to the liquid is yQ(L1z + 'Ihi).1:he po~er;equired ~~~~~ta~ bine delivers is less than this, depending on both its hydraulic and mechanical
pump is greater than t~is, de~ending on the efficiency o t e pump. losses. The head under which the turbine operates is
pumping head hp for this case is
(8.85)
( (8.83)
Here "'i,hL is the loss of head in the supply line plus the submerged discharge loss
(pipe friction and minor losses), but it does not include the head loss in the draft
If the pump discharges a stream through a nozzle, a~ in ~ig. 8.25, not only
tube (DE in Fig. 8.26), since the draft tube is considered an integral part of the
does it lift the liquid a height L1z, but it also imparts a kinetic energy head of
turbine. A draft tube has a gradually increasing cross-sectional area, which re-
duces the velocity at discharge. This enhances the efficiency of the turbine be-
cause it reduces the head loss at discharge (Secs. 8.22 and 16.8). Note that the h,
of Eq. (8.85) represents the energy head the turbine removes from the fluid; this,
, ·· rr- ---!::1
EL: __ ~._:!_Jj of course, is the same as the energy head the fluid transfers to the turbine.
HGL

A
- --- --- ---·1-
l
!
i
i
Ii,
l
i
I
I

F
~ --· --·· --· --------·

Figure 8.25 Figure 8.26 E

Pipeline with pump and nozzle. Pipeline with turbine.
 8.28 Pipeline with Pump or Turbine 325
CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
324 Solving this by trial and error (see Sample Probs. 3.5 and 5.8) or by equation/
polynomial solver (Appendix D), Q = 3.47 cfs
· k and pipelines of Fig. S8.13 have
SAMPLE PR~B~EM 8.13
The ta~ s, {:~:~~trance from the pressure tank is From Eq. (B.7) the other two roots are Q = -2.48 and -0.987 cfs.
the charactenst~cs noted: The suct10e~ tank is submerged. If the pump p puts
flush , and the ~1sc~arge mto th ~ opth flow rate and (b) find the pressure in the So Q = 3.47 cfs ANS
2.0 hp into the liquid, (a) determine e (b) To obtain the pressure p 2 at the suction side of the pump,
pipe on the suction side of the pump.
3.47
¼ = _ = 17.68 fps
0 1963
/ s psi
- 200 ft
= Bin Writing energy Eq. (5.28) from 1 to 2,
Air = 0.030 2 2
= 0.025 :___ _ _ _ _ _ _T
5(144) 17.68 ( 50 ) 17.68 = p2 + + 17.682
10 ft ~ 05 0 025
+ O + O - · 2(32.2) - · 6/12 2(32.2) 'Y
15
2(32.2)
y = 52pcf ;I/
52
from which p2 = -20.6 ft; p2 -20.6( ) -7.43 psi ANS
'Y 144
Figure SS.13 From facing inside cover:

Solution This is equivalent t9 7.43(i21.lo) = 15.13 inHg vacuum.

p - yQhP = 2.0 Note: We should check the absolute pressure against the vapor pressure of the
Eq. (5.40): - 550 550 liquid (Sec. 5.10) to ensure that vaporization does not occur at point 2.
l__ . . . • . -- _,..,.,,,......__........-... . . . .
~ = - : - . , ~ ~

Thus

Figure 8.13 for flush entrance: ke = 0. 5 .. EXERCISES

Writing energy Eq. (5.27) from 1 ( datum) to 3, 8.28.1 An SO-in-diameter pipe (f = 0.025) 7252 ft long delivers water to a powerhouse
at a point 1500 ft lower in elevation than the water surface at intake. When the
2 ( 50 ) v?6 0030 (200) -vf - -vf flow is 450 cfs, what is the horsepower delivered to the plant?
_(__) + O + O + hp - 0.5 ¼g - 0.025 -6/12 -2g
5 144 -
·
-
8/12 2g 2g
52 2 8.28.2 A 10-in-diameter pipeline (f = 0.020) is 3 miles long (Fig. X8.28.2). When
O- 10 + 0 pumping 4 cfs of water through it, with a total actual lift of 25 ft, how much
power is required? The pump efficiency is 72%.
Q Q
and Vs - - -
Eq. (4.6): 0.1963 0.349 ~
Llz
. . f h v,; v; and g = 32 _2 the energy equation reduces to - - ----------- _.1_
Subst1tutmg or P' 6, s, '
21.2 Q2
23.8 + Q - 2.48 = O

Figure XS.28.2
which we cap rewrite as the cubic expression
8.28.3 A 250-mm-diameter pipeline (f = 0.025) is 4.7 km long (Fig. X8.28.2). When
2.48Q 3 - 23.8Q - 21.2 = 0
pumping 100 Lis of water through it, with a total actual lift of 10.5 m, how
much power is required? The pump efficiency is 75%.
8.28.4 If in Sample Prob. 8.13 the vapor pressure of the liquid is 1.9 psia and the
Hffi1: Programmed computing aids (Append_ix C) could help solve problems marked atmospheric pressure is 14.5 psia, what is the maximum theoretical flow rate?
with this icon.
326 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
8.29 Branching Pipes 327
8.28.5 In Fig. 8.26 assume the pipe diameter is 12 in,f = 0.021, BC= 200 ft, and L1z =
120 ft. The entrance to the pipe at the intake is flush with the wall , and discharge
losses are negligible. (a) If Q = 8 cfs of water, what head docs it supply to the
turbine? (b) What power does the turbine deliver if its efficiency is 75 %? A
8.28.6 In Fig. 8.26 assume the pipe diameter is 300 mm,f = 0.021, BC= 60 m, and Liz=
36.5 m. The entrance to the pipe at the intake is flush with the wall, and discharge
losses arc negligible. (a) If Q = 225 Us of water, what head docs it supply to the
turbine? (b) What power docs the turbine deliver if its efficiency is 75%?
8.28.7 A 12-in-diameter pipe 9200 ft long (f = 0.024) discharges freely into the air at
an elevation 18 ft below the surface of the water at intake (Fig. X8.28.7) . It is
necessary to double the flow by inserting a pump. lf the efficiency of the pump Figure 8.27
is 73% , how much power will be required? Branching pipes.

junction equals the flow leaving it and that th T)

shall represent schematically by th . . e pressure head at f (which we
with water at elevation p~·s comm e Itmagllm~ry open piezometer tube shown

~
on o a pipes.) . '
A s there are no pum s the elevation of p ~ .
reservoirs A and C. If pis lev 1with th f st he b~tween the surfaces of
both zero. If pis above the surface of\:~:t::irofBrethservoir B then h2 and Q2 are
and Q - Q + Q If • en water must flow · t B
Figure XS.28. 7 B
out of -and O _;_- Q P~sQbel~wfthe shurf~ce o! reservoir B then the flow m1:~be
f~ ~1 • 2 3• o or t e s1tuat10n shown in Fig ~ w h h
yg gou@rR11+g--C.Qnd.it.ious: · · e ave t e
8.28.8 A 300-mm-diameter pipe 3400 m long (f = 0.022) discharges freely into the air
at an elevation 5.6 m below the surface of the water at intake (Fig. X8.28.7). It 1. Ql = Q2 + Q3, ~
is necessary to double the flow by inserting a pump. If the efficiency of the
pump is 76%, how much power will be required? 2. Elevation of pis common to all three pipes. AA f)" '
t.,f'v\~

8.28.9 Refer to Fig. 8.25. When the pump is delivering 1.2 cfs of water, a pressure gage at The diagram suggests several different r bl j ~
discuss ~ using different meth d pf o 1e~~n~} cases, three of which we will
D reads 25 psi, while a vacuum gage at C reads 10 inHg. The pressure gage at D is o so so ut10)
2 ft higher than the vacuum gage at C. The pipe diameters are 4 in for the suction
pipe and 3 in for the discharge pipe. Find the power delivered to the water. Rigorous Solutions
8.28.10 Refer to Fig. 8.25. When the pump is delivering 35 Lis of water, a pressure gage
at D reads 175 kPa, while a vacuum gage at C reads 250 mmHg. The pressure . When we know the pipe wall t . 1
gage at Dis 600 mm higher than the vacuum gage at C. The pipe diameters are and we know that the friction ~:c~~tJ we_ can ~~imate its e value (Table 8.1),
100 mm for the suction pipe and 75 mm for the discharge pipe. Find the power Reynolds number of the flow B vanes wit the e/D of the pipe and the
. ecause we are not consider· . 1
delivered to the water. can use the equations and methods f S m_g mmor asses, we
basic scientific calculator we can ol e~s. 81_.12-8.17. In particular, using only a
,
th e H aa1and equation (8.52) so ve pipe mes for h (Type 1 bl ) .
with E (8 B)· pro ems usmg
8.29 problems) using Eq (8 56)· d q. . 'we can solve for V or Q (Type 2
BRANCHING PIPES SkAWvl 1ems ) using Eqs. (8.57)-(8
· · , an more rarely we can sol f D (T
62) We f ' . ve or ype 3 prob-
For convenience,Qet us consider three pipes connected to three reservoirs as)in trial and error which can be . pr_e er these equat10ns because they avoid
. ' come qmte confusing when b.· d
Fig. 8.27 land connected together or branching at the common junction point J. tna1-and-error techniques needed t
Th .
1 f b .
o so ve or ranchmg flow
com me with other
Actually~e can consider that any of the pipes is connected to some other des- e followmg three cases illustrate some of h " . " .
tination than a reservoir by simply replacing the reservoir with a piezometer methods used to solve the different t f t e manual tnal-and-error
tube in which the water level is the same as th~ reservoir surface(We shall sup- case all the required pipe data (len t?pe~. o problems that ca_n occur. In each
pose that all the pipes are sufficiently long that we can neglect' minor losses and are known. g s, iameters, and matenals fore values)
velof~!Y heads, so hL = h1, which we shall designate ash )
'07e name the pipes and flows and corresponding fnction losses as in the di- Case 1. F'm d th e fl ow to or from two reserv . d .
agram. The continuity and energy equations require that the flow entering the one of these, given the other fl d Oir_s, an the surface elevat10n of
ow an two elevations and all the pipe data. . -
 8.29 Branching Pipes 329
328 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
Q1 > Q?, then we must raise p to satisfy continuity atl caus· t .
60 1Elevation of P reservoir B , and we shall have Q = Q + Q . 'f Q ' mg wa er tto flow mto
3
satisfy continuity at J, causing w~ter t~ flow ~~t of ~s~:~:;~w:n~uS l~w~\; to
Q1 + Q2 = Q3, From here on the solution proceeds by adjusti~g p :ef~raCas:v2~

Note: For any of the above three cases . .

er~or by setting up the governing equations ~n~\~f~nav~~d manual tnal and
45 usmg equation solving software like Mathcad or Excel (s!e Aem si;ulta~eously
and D.2-D.3). There will be the usual four overni . ppen ixes -~-C.3
40 flow continuity equation for the junction an~ d n~_equations for eachhne, a
- 8 - 6 - 4 -2 0 2 4 6 8 one or two equations relating the variods he~d ~Es::s i~i~: :~e case adressed,
Figure 8.28
~Q at] to solve for, the success of the procedure becomes mo. many un nowns ·r
values, and we may have to tr differ t re sensi ive to the guessed
We can solve this problem directly, without trial and error. Suppose that Q 1 ;~~;~ ~r~:i~!is so straightiorward~~t rs~;l~:t~a~;; i!;:~:ti)~~~i: ~~:: !: :~
and the elevations of A and B are given. Then we can directly determine the
head loss h (Type 1), using Fig. 8.11 or Eq. (8.52) to find the proper value off. ~fti11¥;,;.;;;_·...r-..i. .......... y:::w~-,.r-- ... *"' ..::.;.~V.:•~ ~.--,.- ~ ~ ; , ; . " . \ . ~ . ~ - ~...'l;,rr;~ ... ~ , - • • .,.1..;",,..C.J\:''d'I"",. ,.·..4!:t*"T#~ t .'l,1f..."=llr.W."..,_.,'.k~t;..:i.'-":i."r-"'.i'ff··_···• , ...

Knowing h 1 fixes P, so now we can easily obtain h 2 . Knowing h 2 enables us to di-

1
rectly determine the flow in pipe 2 using the Type 2 equation (8.56). Condition 1
SAMPLE PROBLEM 8.14 Given that in F 8 27 1. . .., .• '"· _·"
(continuity at junction J) then determines Q 3, which in turn enables us to di-
diameter, pipe 2 is 1500 ft of 10 in dia it : , I? Pe 1 1s 6000 ft of 15 m 't
all asphalt-dipped cast iron. The elevm~ er, an pipe 3 is 4500 ft o~ 8 in diameter,
rectly find h (Type 1), in the same way as for line 1. Finally, P and h 3 define the and Care 250 ft and 160 ft respective~~o;sdo~h~.w~er surfaces m reservoirs A
3
required surface elevation of C. reservo)r Bis 3.3 cfs. Finl the surface ~le~ . e isc arge {_22 of 60°F water into
basic scientific calculator' (b) using equat1·oatlonl <?f resefrvo1r
n so vmg so tware.B: (a) usmg only a
Case 2. Find the flow to or from tw·o reservoirs, and the surface elevation of
the third, given the other flow and two elevations and all the pipe data. Solution
Let us suppose that Q 2 and the surface elevations of A and Care given.6 This is a Case 2 problem.
Then we knqw h + h = L'.1h 13 , say. We may use various solution approaches;2
1 3 Table Al for water at 60oF: v = 12.17 x 10-6 ft2/sec.
we shall qiscuss a more convenient one. In this, we assume the elevation of P,
which yields values for h and h 3 , and so Q1 and Q 3 via Eq. (8.56). If these do not Pipe: 1 2 3
1
satisfy the discharge relation at J then we must adjust P until they do. To help us
converge on the correct elevation of P, we can plot the results of each assump-
I L, ft 6000 1500 4500
tion on a graph like Fig. 8.28. For ::2:Q at J, inflows to J are taken as positive and
f D, ft 1.25 10/12 8/12
ij e, ft (Table 8.1) 0.0004 0.0004 0.0004
~
outflows as negative. Two or three points, with one fair!Y close to the vertical
axis, determine a near~raight) curve that intersects that axis at the equilibrium L/D 4800 1800 6750
leve o , where ::2:Q = 0, as required. Last, we can determine h 2 from Q2 and ( A = nD2/4, ft 2 1.227 0.545 0.349
Eqs. (8.52) and (8.13), and find the required surface elevation of B. i~ e/D 0.00032 0.00048 0.0006

Case 3. Find the flow in each pipe, given the surface elevation of all three t
~
(a) Find the elevation of p by trial and error.
reservoirs· and all the pipe data.
This is the classic three-reservoir problem, and it differs from the foregoing I
~ Elevation of plies
turbulent flow. between 160 an d 250 f t. Calculate V from Eq. (8.56) assumin_g
Trials:
cases in that it is not immediately evident whether the flow is into or out of reser- r,.

voir B. We can readily determine this direction by first assuming no flow in pipe 2;
that is, assume the piezometer level P at the elevation of the surface of B. The
head losses h and h then determine the flows Q1 and Q3 via Eq. (8.56). If
1 3
'
~
~·
Elev. P h1

200
230
50
20
h3
40
70
Yi
6.444
4.013
½
4.481
5.984
662,000
412,000
R1 R2
245,000
328,000
Ql
7.907 1.564
4.925 2.088
Q3 LQ
+3.04 Up
0.463 Down
MoveP?

26Other approaches include (a) assuming distributions of the flows Q 1 and Q 3, knowing
that Q - Q = Q , and (b) by substituting for the h.~ in h 1 + h 3 = '1h 13 using Eq. (8.13)
3
with ½1 written
2
in terms of Q 3 , and ½ written in ten~ of Q 2 + Q 3 , and successively
fffiil: this
with Programmed
icon. computing a·d
1 s (A ppen d.ix C) could help solve problems marked

solving the resulting quadratic equation for Q 3 whil~,.converging on f values.

330 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
8.29 Branching Pipes 331
The tabulated pipe data are the same as for Sample Prob. 8.14.
Interpolation (Fig. 8.28): (230 - Elev. P)/(230 - 200) = 0.463/(0.463 + 3.04);
Elev. P = 226.03. (a)· Trial 1. First, try pat elevation of reservoir surface B = 500 ft:
LQ Move P?
Pipe: 1 2 3
226 24 66 4.412 5.805 453,000 318,000 5.414 2.026 +0.088 Up
h, ft 25 0 70
Interpolation, (Fig. 8.28): (230 - Elev. P)/(239 - 226) = 0.463/(0.463 + 0.088);
Elev. P = 22b.64.
Close enough! Note: These adjustments are very suitable for making on a
I V2gDh/L, fps
V, fps (Eq. 8.56)
Q = AV, cfs
0.579
4.51
5.53
0
0
0
0.817
5.98
2.09
spreadsheet (Appendix C.2).

2
Q
A2
3.3
0.545 . '
D 2 V2
V: = - 2 = - - = 6055fps· R2 = - - = 416500
V '
I
! At I, LQ ~ inflow - outflow= 5.53 _ 2 09 = 3
B. ANS 3 '
.
I must be raised ( to reduce Q and increase .Q )· th .44 cfs. Thi_s must ?e zero, so p
t i en water will flow mto reservoir
All three R values are turbulent, so the use of Eq. (8.56) and these results are
(b) Trial 2. Raise P. 500 ft< Elev· p < 525 ft · Try p ate 1evatlon
. 510 ft:
valid.
Eq. (8.52):
Ji = 0.017 61; Eq. (8.13): h 2 = 18.05 ft
h, ft 15 10 80
Elev. B = Elev. P - h2 = 226.64 - 18.05 = 208.59 ft ANS V2gDh/L, fps 0.449 0.598 0.874
(b) Using equation solving software such as Excel or Mathcad (Appendixes C.2- V, fps (Eq. 8.56) 3.46 4.49 6.41
C.3), we note that there are 14 governing equations. Two of these are Check R 355,000 307,000 351,000
Q = AV, cfs 4.24 2.42 2.24
h + h 3 = 250 - 160 = 90, Q 1 = Q 2 + Q3
1
The remaining 12 equations are the four governing equations (Sec. 8.14) for · = -042
Atl, LQ = 4.24 - 2.42 - 224 · cfs. By mterpolation
. . (using Fig. 8.28),
each of the three pipes. 1,
The complete output for the Excel and Mathcad solutions is provided in i 510 - Elev. p 0.42
510 - 500 = 0.42 + 3.44 , Elev. P = 508.91 ft

IJ
Appendixes D.2 and D.3, respectively.
The solver provides values for the 14 unknowns (three values each for f, V, R ,
h, and two values for Q) from which it calculates
Elev. B = 208.53 ft ANS h, ft 16.1 8.9 78.9
Note: Observe how simple and clear the Mathcad solution is. Vr=-2g---:D=-=h-.,,/L-, fps 0.465 0.564 0.868
V, fps (Eq. 8.56) 3.59 4.19 6.36
Check R 339,000 287,000 348,000
Q = AV,cfs 4.40 2.28 2.22 ANS
SAMPLE PROBLEM 8.15 With the sizes, lengths, and material of pipes given in
Sample Prob. 8.14, suppose that the surface elevations of reservoirs A, B, and C
are 525 ft, 500 ft, and 430 ft, respectively. (a) Does water enter or leave reservoir
B? (b) Find the flow rates of 60°F water in each pipe. Use only a basic scientific
calculator.
Solution
This is a Case 3 problem. Find the elevation of P by trial and error.
6 2 Nonrigorous Solutions
Table A.1 for water at 60°F: v = 12.17 x 10- ft /sec.
If the ~alue of the friction factor ( constant f C . ' . .
each pipeline we must use th . ' mv, or Mannmg s n) is gzven for
Then, to solv~ the three cases -~s;o~ngorous head-loss equations of Sec. 8.19.
liffll: Programmed computing aids (Appendix C) could help solve problems marked for the rigorous solutions bu/ inst~:~us;ed: we hfollow e~actly the same steps as
, 0 usmg t e equations of Secs. 8.12-8.17
'
with this icon.
332 CHAPTER 8: Steady Incompressible Flow ill Pressure Conduits
8.30 Pipes in Series 333
we simply use Eqs. (8.65)-(8.69). Of course, we must first determine the appro-
priate Kand n value for each pipeline, and select accordingly from Eqs. (8.66)- AtJ, ~Q = 0.085~ - 0.058~ - 0.0450 = -0.0184 m3!s. This-must, be zero sow
mu~t lower P. By mterpolation (Fig. 8.28), ' e
(8.68). Notice; however, that we can easily solve Case 2 directly if n = 2 (Darcy-
Weisbach or Manning) by writing the known elevation difference '1h 13 = h 1 + 155 - Elev. P 0.0184
h3 = K 1Qf + K 3 Q} = K 1(Q 2 + Q 3)2 + K 3 Q}, which is a quadratic equation in Q 3, 155 - 150 0.0184 + 0.0822 ' Elev. p = 154.09 m
the only unknown.
Trial 3. Try Pat elevation 154 m:
,.
h,m 6 4 34
Q, m3ts (Eq. 8.69) 0.0938
SAMPLE PROBLEM 8.16In Fig. 8.27 pipe 1 is 300 mm diameter and 900 m 0.0518 0.0443 ANS
V, mis (Eq. 4.7) 1.327
long, pipe 2 is 200 mm diameter and 250 m long, and pip~ 3 is_ 150 mm diameter 1.649 2.51
and 700 m long. The Hazen-Williams coefficient for all pipes 1s 120. The su~face
elevations of reservoirs A, B, and Care 160 m, 150 m, and 120 m, respectively. AtJ, LQ = -0.0023 m3!s. This is close enough.
(a) Does water enter or leave reservoir B? (b) Find the flow rate in each pipe. All Vs are < 3 m:s (Sec. 8.18), so these solutions using Eq.· (8.67) are valid.
Use only a basic scientific calculator. Note: T~ese adJustments are very suitable for making on a spreadsheet
(Appendix C.2).
Solution
This is a Case 3 problem. Find the elevation of P by trial and error. Use the
Hazen-Williams form (8.67) of Eq. (8.65) in SI units:

l0.675L 8.30 PIPES IN SERIES

K= cM?D4.87, n = 1.852
The discus~ion. in ~ec. 8.27 addressed the case of a single pipe of constant diam-
eter. _I~ a pipelme 1~ made up o_f l~ngths of different diameters, as in Fig. 8.29,
Pipe: 1 2 3
cond1t10ns must satisfy the contmu1ty and energy equations, namely:
L,m 900 250 700
D,m 0.3 0.2 0.15 \
L/D 3000 1250 4667 (8.86)
CHW 120 120 120
n 1.852 1.852 1.852 (8.87)
K 480 961 10924
If we are given the_rate of discharge Q, the problem is straightforward. We
T?ay find _the head loss dJrectly by adding the contributions from the various sec-
(a) Trial 1. First, try Pat elevation of reservoir surface B = 150 m: ~10ns, as m_Eq. (8.87); __as discussed in Sec. 8.14, hL = L\Q. If we are given empir-
_1cal coefficients ?r constant f values, we can do the same thing using Eq. (8.65)
h,m 10 0 30 and the appropnate values of Kand n selected from (8 66)-(8 68) If h
Q, m3/s (Eq. 8.69) 0.1236 0 0.0414 w · h · . · · . , owever,
·'· · _, e are g1~en t e pipe matenal ore, the result will be more accurate, because it
~se~ ~he ngorous Darcy-~eis~ach approach. Then we use Eq. (8.13) to find the
At J, LQ = inflow - outflow = 0.1236 - 0.0414 = 0.0822 m3/s. T~is mus~ be md1VIdual head loss contnbutions after finding e/D, V, R, and f for each pipe.
zero, so we must raise P (to reduce Q 1 and increase Q 3), then water will flow into
reservoir B. ANS
(b) Trial 2. Raise P. 150 m < Elev. P < 160 m, so try Pat elevation 155 m:

h,m 5 5 35
Q, m3/s (Eq. 8.69) 0.0850 0.0584 0.0450

IHI: Programmed computing aids (Appendix C) could help solve problems marked Figure 8.29
with this icon. Pipes in series.
 8.30 Pipes in Series 335
334 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits

If we are given the total head loss hL and want to find the flow, the pro·01em Pipe: 1 2 3
is a little more involved. L,m 300 150 250
Using the nonrigorous equations, we again substitute Eq. (8.65) into Eq.
D,m 0.3 0.2 0.25
(8.87), to get e/D 0.000833 0.00125 0.00100
hL = hf = KiQ1 + K2Q2 + K 3Q'; + · · · fmin (Fig. 8.11) 0.019 0.021 0.020

But since all the Qs are equal from Eq. (8.86), this becomes
Assuming these friction factor values
(8.88) '
hL = (Ki+ K2 + K3 + · · ·)Qn = (~K)Qn
L1z = hL = ht = 10 = o.019(300)v;2 + o.021(150)½2 (250)½2
So, knowing the pipe information and which nonrigorous equation we must use, \ 0.3 2g 0.2 2g + 0.020 0.25 2g
we can solve for Q. Last, we must check that all velocities are in the required
ranges for the empirical equations to be valid (Sec. 8.18).
If we wish to use the more accurate, rigorous Darcy-Weisbach approach to From continuity, ~ = i (~J = i (~:~J 5 _06 ~: .
find Q, we must note that in Eq. (8.88) each K has now become a function of a dif-
ferent f. The preferred manual method of solution is similar to that just discussed, V32 = 2
and we call it the equivalent-velocity-head method. Substituting from Eq. (8.13) · Similarly, 2.07½
2g . 2g
into Eq. (8.87) and including minor losses if we wish (usually if L/D < 1000),
2
- + L k )Vi
)Vi2 + ( rL2
Li + L k - l O -_ ½ ( 0.019-
1000 + 0.021-5.06
750 1000 2 07)
+ 0.020--
h = f- - + ··· and thus
2g 1 1 1 ·
L ( J 1 D1 1 2g D2
J2 2 2g
Using continuity Eq. (4.17), we know Df V1 = Df½ = D~½, etc., from which we ½2
can express all the velocities in terms of one chosen velocity. So, by assuming from which g = 0.0713 m
2
reasonable values for eachf (e.g., from Eq. (8.54) or Fig. 8.11), for any pipeline,
however complex, we can write the total head loss as So Vi = V2(9.81 m/s 2)(0.0713 m) = 1.183 mis
hL =
v2
K- The c<?rresponding values of Rare 0.31 x 106 o 47 106
't~nd 0.37 x ~~ ; the
(8.89) 6
2g correspon~mg friction factors are only slightly diff~r~nt;
ass_umed, smce the flow occurs at Reynold b rom ose we ongmally
where Vis the chosen velocity. We can solve this equation for the chosen V, and pipe flow. So s num ers very close to fully-rough-
so can obtain the V and R and f values for each pipe. For better accuracy, we
should replace the assumed values off by the values just obtained, and obtain Q = Ai½ = ¼n-(0.30)21.183 = 0.0836 m 3/s ANS ;
an improved solution. When the f values converge V is correct, and we can
Note: We would have obtained greater a "f .
factors to match the pipe-friction ch ;curacy\ wel had adJusted the friction
calculate Q. culated them by Eq. (8.5 ) and .f a~ g1.ore cdose Y. (further trials) or cal-
Q = 0.0821 m3/s. 2 ' 1 we a me1u ed mmor losses. In that case

SAMPLE PROBLEM 8.17 Suppose in Fig. 8.29 the pipes 1, 2, and 3 are 300 m
of 300 mm diameter, 150 m of 200 mm diameter, and 250 m of 250 mm diameter,
respectively, of new cast iron and are conveying 15°C water. If .dz = 10 m, find
the rate of flow from A to B using only a basic scientific calculator. Neglect
We can avoid manual iteration for f b l · ·
minor losses.
and D.2-D 3) Th .
sir
using equation solving software like Mathc!ct v-l{1x~:/m(Aultane~~s eqCuations
ppenu1xes .2-C.3
Solution . ·.f . ere are the usual four equat10ns for each pipe (Sec 8 14) plus
Table 8.1 for cast-iron pipe:e = 0.25 mm = 0.00025 m. Eq · (8·87) , 1 necessary we may ·1 f . · · .,
6 2 loss equations with the form of E;a(~ ~O)c~~u~~ ~~ mm~r losses _by using head
Table A.1 for water at l5°C: V = 1.139 X 10- m /s . . ample, there are therefore 13 s· . It. . r e . ree p1p~s of Fig. 8.29, for ex-
the usual manner for 13 u k imu (aneot equations, which we may solve in
D.2-D.3) The n no~ns see ample Prob. 8.14b and Appendixes
three val~es eac~:;,~w~ ;e ed1thfer the flow rate or the total headJ_9ss, and
liffll: Programmed computing aids (Appendix C) could help solve problems marked L~ , , an ..
with this icon.
 ·bte Flow in Pressure Conduits
8.31 Pipes in Para.,/el 337
336 CHAPTER 8: Steady Jncompressi
function of a different f The preferred manual method of solution is similar to
the preceding. Writing Eq. (8.13) for each line, including minor losses if we wish,
8.31 PIPES IN PARALLEL ··
Inth e case of fl
ow through two or more parallel pipes, ~sin Fig. 8.30, conditions
. . d gy equations: L
L + ~k -
h = ( f-
D 2g
)V 2

must satisfy the following contmmty an ener

(8.90) where ~k is the sum of the minor-loss coefficients, which we can usually neglect
Q = Q1 + Q2 + Q3
if the pipe is longer than 1000 diameters. Solving for Vand then Q, we obtain the
(8.91)
hL = hLl = h[,2 = hL3 following for pipe 1:

because the pressures a t A an

d B are common to all pipes. Problems may be
posed in various ways. . . roblem is straightforward. We ca~ find tl~e
If the head loss hL is given, th~ p t ·butions from the various pipes, as m
total discharge directly by addi1:~ t le con ~~ess coefficients or constant f values,
e given empmca roug · l f K and
Eq. (8. 90) . If we ar . . E . (8.69) and the appropri~te va ues_o
we can do the same thmg u mg q s given the pipe matenal ore, the
selected from (8.66)-(8.68). If, howeve~t'· we artehe rigorous Darcy-Weisbach ap-
urate because i uses . (
result will be more ace ? de endent Type 2 problem for each pipe see
proach. Then we h_ave an m P . tl b E . (8.56) for example. . .
Secs. 8.14-8.17), which we can solvQe dir~cw~nito rind the head loss and indiv1d-
If we are given the total flow a_n
· rttle more involved. . b
ual flows , the proble~ is a i . d neglecting minor losses, we agam su -
Using the empmcal equations, an
Q, =A,½= A,
,JNLi/D,)
2ghL , ;,"
+ », = c,vhL (8.93)
where C1 is constant for the given pipe, except for the change in/with Reynolds
number. We can similarly express the flows in the other pipes, using assumed
reasonable values off from Fig. 8.11 or Eq. (8.54). Finally, Eq. (8.90) becomes
n
Q = C1-v,;;_,_ + C2-v,;;_,_ + C3-v,;;_,_ = (C1 + C2 ,+ C3)-v,;;_,_ (8.94)
This enables us to find a first estimate of hL and the distribution of flows and ve-
locities in the pipes. Using these, we can-next make improvements to the values
off, if indicated, and if necessary repeat them, until we finally obtain a correct
determination of hL and the distribution of flows.
If we use the turbulent-flow equation (8.51) or (8.52) to obtain f, we must
remember to confirm that the Reynolds number is in the turbulent range. We

Q = r~:r c~:r (tr

stitute Eq. (8.69) into Eq. (8.90), to get

+ + + ---
But since all the hr s(= hLs) are equal from Eq. (8.91), this becom;s 1/•
can precheck the likelihood of laminar flow occurring in any of the pipes by cal-
culating an "average" flow velocity from the total flow divided by the total area
of all the pipes, and using this velocity to obtain an indicator R for each pipe.
r _,-T?I ,__,,._ . _,___,..,....,,,_....,.,,..,,W,,.,,.,,._-=-..,.._..,.._,,....,,....,,_. ..,

Q ~ (hr)'l•lGJ + (~2r + (~J" + · ·1 = (hr)'/n :2:(K) (8.92)

~
IBIIIJ
SAMPLE PROBLEM 8.18 Three pipes A, B, and Care interconnected as in
Fig. S8.18. The pipe characteristics are as follows:
. d he empirical equation we must use, we
So, knowing the pipe informati~nt~~ intdividual flows using Eq. (8.6~): Last, we Pipe D (in) L (ft) f
can solve for hr We can t~~n fin . th required ranges for the empmcal equa-
must check that all velocities are m e A 6 2000 0.020
B 4 1600 0.032
tions to be valid (Sec. 8.18). . Dare -Weisbach approach to find hL and
If we wish to use the ngorous .
the individual Qs, we must note that m q. .
J
(8 92) each K has now become a
C 8 4000 0.024

-El. 200 ft
A

Figure S8.18

liffil: Programmed computing aids (Appendix C) could help solve problems marked
Figure 8.3._0 with this icon.
Pipes in parallel.
 8.32 Pipe Networks 339
. . ·b/e Flmv in Pressure Conduits
CHAPTER 8: Steady Jncomp1 ess1 Pp Vt v,2
338 Check: 120 + - - 144- = 50 + ~
· . h . flow m
t will . each pi·pe . Find also the pressure. at 'Y 2g 2g
Find the rate at wh1c wa er h t than 1000 diameters, so neglect mmor
point P. All pipe lengths are muc grea er Pp ( 5.77) 2
145---70 5.01 ft
losses. 'Y 2(32.2)
Solution . ~2 4000 VJ Vc2 , So pp/'Y = 5.01 ft and PP= (62.4/144)5.01 = 2.17 psi. ANS
2000 ~ - 0 024-- - = 0 + 50 + 2g
Eq. (5.28): o + 200 + 0 - 0.020 6/12 2g · 8/12 2g Note: In this example we were given the values off for each pipe as known.
Actually f depends on R [Fig. 8.11 or Eq. (8.52)]. Usually we know or assume
v? VJ (1) the absolute roughness e of each pipe, and achieve an accurate solution by trial
150 = 80~ + 145-
2g 2g
and error until the fs and Rs for each pipe have converged.
i.e.,

62¼ + 42"\'8 82Vc

Continuity:
(2)
36¼ + 16"\'8 = 64Vc We can avoid manual iteration for f by solving simultaneous equations
i.e., using equation solving software as in Mathcad or Excel (Appendixes C.2-C.3
2000 V} 1600 Vl and D.2-D.3). There are the usual four equations for each pipe (Sec. 8.14), plus
h 0020 -- = 0032 I
Also, hfA = fB = · 6/12 2g . 4 12 2g Eq. (8.90); if necessary, we may easily account for minor losses by using head
loss equations with the form of Eq. (8.80). For the three pipes of Fig. 8.30, for ex-
80V} = 153.6VJ, VB = 0. 722 ¼ ample, there are therefore 13 simultaneous equations, which we may solve in
i.e., the usual manner for 13 unknowns (see Sample Prob. 8.14b and Appendixes
· Substituting into (2): D.2-D.3). The unknowns are either the head loss or the total flow rate, and
36¼ + 16(0.722¼) = 64Vc three values each of Q, V, R, and f.
It is instructive to compare the solution methods for pipes in parallel with
47.5¼ = 64Vc, ¼ = 1.346Vc
those for pipes in series. The role of the head loss in one case becomes that of
the discharge rate in the other, and vice versa. Students should be already fa-
Substituting into (1):
miliar with this situation from the elementary theory of de circuits. The flow cor-
(l.346Vc) + 145~
2 v?
= 289.9-
VJ responds to the electrical current, the head loss to the voltage drop, and the
150 = 80 2g 2g 2g frictional resistance to the ohmic resistance. The outstanding deficiency in this
analogy occurs in the variation of potential drop with flow, which is with the first
VJ = 2(32.2)150/289.9 = 33. 3 power in the electrical case (V = IR) and with the second power in the hydraulic
ANS case (hL ex V 2 ex Q 2) for fully developed turbulent flow.
Ve = 5.77 fps, Q c -_ A c Vrc = (0 349)5.77
·
= 2.01 cfs
ANS
¼ = 1.346\f = 7.77 fps, QA = (0.1963)7.77 = 1.526 cfs
8.32 PIPE NETWORKS
VB = 0.722¼ = 5.61 fps
In municipal distribution systems, pipes are frequently interconnected so that
_ A -v; = (0 0873.)5.61 = 0.489 cfs ANS
QB - B B .
the flow to a given outlet may come by several different paths, as in Fig. 8.31. As
a result, we often cannot tell by inspection which way the flow travels! as in pipe
As a check, note that QA
+ Q~ = Qc is satisfied. BE. Nevertheless, the flow in any network, however complicated, must satisfy
To find the pressure at P. the basic relations of continuity and energy as follows:

¼_2 1. The flow into any junction must equal the flow out of it.
Eq. (5.28):
0 + 200 + 0 - 80-
2g
= PP
-'Y + 120 2. The flow in each pipe must satisfy the pipe-friction laws for flow in a
single pipe.
2 3. The algebraic sum of the head losses around any closed loop must be
PP (7 .77 ) = 5.01 ft
zero.
'Y = 80 - 80 2(32.2)
340 CHAPTER 8: Steady Incompressible Flow in Pressure Conduits
Figure 8.31
Pipe network.
Most pipe networks are too complicated to solve analytically by hand using
rigorous (variable f) equations, as _was possible in the simrler cases of parallel
pipes (Sec. 8.31). Nowadays they are readily solved by specially developed com-
puter programs (Appendix C.5). However, in many case~ we ~annot ~r~dict the
capacity requirements of water distribution systems wit~ high prec1s~on, and
flows in them vary considerably throughout the day, so high ~ccuracy m c~lcu-
lating their flows is not important. As a result, the use of nonngorous equati~ns
(Secs. 8.18-8.19) are very acceptable for this purpose. The method of successive
approximations, due to Cross, 27 is such a method that was popular before t~e
advent of computers. We will review it here to help students underst~nd the flmd
mechanics of pipe networks and evaluate computer-generated solutions. It con-
sists of the following elements, in order:
Step J: By careful inspection assume the most reasonable distribution of
flows that satisfies condition 1.
Step 2: Write condition 2 for each pipe in the form
(8.95)
where Kand n are constants for each pipe as described in Sec. 8.19. If minor
losses are important include them as in Eq. (8.93), ':"h~ch yield_s K = l/C 2 and
n = 2 for constant f We may include minor losses w1thm any pipe or loop, but
must neglect them at the junction points.
Step J: To investigate condition 3, compute the algebrafc sum of the head
losses around each elementary loop, 22hL = 22KQn. Consider losses from clock-
wise flows as positive, counterclockwise negative. Only by good luck will these
add up to zero on the first trial.
Step 4: Adjust the flow in each loop by a correcti~n L1Q to b~lan~e the
head in that loop and give 22KQn = 0. The heart of this method hes m the
27 H.Cross, Analysis of Flow in Networks of Conduits or Conductors, Univ. III. Eng.
Expt. Sta. Bull. 286, 1936.
8.32 Pipe Networks 341
following determination of L1Q. For any pipe, we may write
Q = Qo + L1Q
~here Q is the correct discharge and Q 0 is the assumed discharge. Then, for each
pipe,
If L1Q is small compared with Q 0 , we may neglect the terms of the binomial se-
ries after the second one, so that
hL = KQo + L1QKnQ3 - 1
For a loop, 2.hL = 2.KQn = 0, so because L1Q is the same for all pipes in that
loop, .
As we must su~ _the corrections of head loss in all pipes arithmetically (treating
all terms as positive), we may solve this equation for L1Q,
L1Q
(8.96)
since, from Eq. (8.95), hJQ = KQn-i_ We emphasize again that we must sum
the numerator of Eq. (8._96) algebraically, with due account of each sign, while
we must s~m th~ denom_mator arithmetically. Note that the Q0 jQ0- 1j in the nu-
merator g1_ves_ this quantity the same sign as the head loss. The negative sign in
Eq. (8.~6) 1~d1ca_tes that when there is an excess of head loss around a loop in the
~lockw1se duection,_ we must subtract the L1Q from clockwise Q values and add
0
it to counterclock~1se ones. The reverse is true if there is a deficiency of head
loss around a loop m the clockwise direction. .
. Step 5: After we have given each loop a first correction, the losses will
shl! not balance, because of the interaction of one loop upon another (pipes
which are common to two loops receive two independent corrections, one for
each loop). So we repeat the procedure, arriving at a second correction and so
on, until the corrections become negligible. '
We may use either form of Eq. (8.96) to find L1Q. As values of K appear in
bot~ the numerator and denominator of the first form, we can use values pro-
por~10nal to the actual K to find the distribution. The second form is more con-
vement for use with pipe-friction diagrams for water pipes.
~n attractive feature of this approximation method is that errors in com-
putation have the same effect as errors in judgment and the process eventually
corrects them.
 8: Steady Incompressible Flo,r in Pressure Conduits
8.33 Further Topics in Pipe Flow 343
342 CHAPTER
As noted earlier, varying demand rates usually make high solution accu-
racy unnecessary with pipe networks. However, if high manual accuracy is re-
SAMPLE PROBLEM 8.19 If the flow into and _out of a t~o-loop pipe system a~e
· p· S8 19 determine the flow m each pipe usmg _only a basic quired for some reason, we can first solve the problem in a similar manner to the
as s h own m 1g. • , df h · preceding example using the Darcy-Weisbach Kin Eq. (8.65) and constant f val-
scientific calculator. The K values for each pipe were calcu;ate ron2_ t e pipe
and minor loss characteristics and from an assumed value 0 f, and n - 2.
1 ues. Then we can use the resulting flows to adjust the f and K values, and repeat
the process (more than once if necessary) to refine the answers. The value of
such refinement is questionable, not only because of uncertainties in the de-
100 cfs 25 cfs mand flows, but also because of uncertainties in thee values (pipe roughness)
K=1 K=4 _/ we must use (see Sec. 8.15). Usually when we adjustfvalues they change by only

□□
II~
a few percent, but we can see in Fig. 8.11 that for smoother pipe it is possible for
f- ~ ~ them to change by as much as a factor of five.
~
We can solve simple networks without approximation and manual itera-
,.;, y / K=5 ' ~ tion by solving simultaneous equations using equation solving software like that
25 cfs 50 cfs in Mathcad and Excel (Appendixes C.2-C.3 and D.2-D.3). For networks con-
Figure S8.19
taining i pipes, Si equations are required if using the Darcy-Weisbach equation
with variable f, and 2i equations are required if using the, simplified Eq. (8.95)
1
Solutt,_; a first step, assume a flow •in e~~h pipe such that continuity holds at all with constant friction factors. These required equations include (a) the usual
·unctions. Take clockwise flows as positive. Calcul~te L1Q f~)f each loop, ma~e (condition 2) flow equations for each pipe (four or one per pipe, depending on
~orrections to the assumed Qs, and repeat several times until the L1Qs are qmte the equations used); (b) flow continuity equations (condition 1) at all but one of
small. the j nodes (as these imply continuity at the last node); (c) equations for the sum
of the head losses around i - j + 1 loops (condition 3). The unknowns we want
Right loop
Left loop 100 25 to find for each pipe are hv Q, V, R , and fusing the Darcy-Weisbach equation,
K = 1K =4 /
1 ~v~ N
nlKQo- 11 or only hL and Q using Eq. (8.95).
hL = KQ 0 n\KQ8 - \ ,r, 10 11 25 11
+~ :.::25 ~ 4 X 502 = 10,000) 4 X 2 X 50 = 400 The pipe-network problem lends itself well to solution by use of a digital
1 X 60 2 = 3,600) 1 X 2 X 60 = 120
2 25 2 = 1,250) 2 X 2 X 25 = 100 !
4 X 10 2 = 400) 4 X 2 X 10 = 80 ~tr /K = 5"-... X computer. Programming takes time and care, but once set up, there is great flex-
25 50
4 X 10 2 = 400 ; 4 X 2 X 10 = 80 ibility and it can save many hours of repetitive labor. Many software packages
3 X 40 2 = 4,800 J 3 X 2 X 40 = 240 First approximation 5 X 25 2 = 3, 125} 5 X 2 X 25 = 250 are now available to simulate water distribution networks; see Appendix C.
800} 440 7,725) 830

-(+7725)
-(-800) = 2) LlQ1 = 830 = 9'
.:1Ql = 440 8.33 FURTHER TOPICS IN PIPE FLOW

25
I The basics of steady incompressible flow in pressure conduits discussed in this
chapter are just an introduction to many more advanced subjects involving flow
After first correction
4X41 2 =6,724) 4 X 2 X 41 = 328 in conduits. Flow through submerged culverts is a special case discussed in
1 X 62 2 = 3,844) 1 X 2 X 62 = 124 2 X 2 X 16 = 64
2 X 162 = 512)
4 X 21 2 = 1,764) 4 X 2 X 21 = 168 Sec. 10.23. Where sewers must dip to flow under obstacles like streams, they
4 X 21 2 = 1,764} 4 X 2 X 21 = 168 flow full and are known as inverted siphons or sag pipes; turbulence, and there-
3 X 38 2 = 4,332} 3 X 2 X 38 = 228 5 X 34 2 = 5,780} 5 X 2 X 34 = 340
1,276) 520
308} 900 fore velocities, must remain adequate when flow rates vary in order to prevent
suspended solids from accumulating at such low points. Later in this text, we
-(+1276) discusss unsteady flow in pipes (Chap. 12), involving both moderate rates of
.:1Q2=~ = 2} change and the very rapid changes associated with water hammer. We often in-
After second correction
vestigate or simulate such flows using numerical methods, usually on a com-
Further corrections can be made if greater accuracy is desired. puter. In Chap. 13 we introduce the analysis of the flow of compressible fluids
(gases), and in Chap. 11 we describe methods ofjlow measurement in pipes.
When we inject a different fluid into a pipeline, the main flow transports it
(advection ), and it also mixes and spreads through the main fluid by the
processes of ionic and molecular diffusion, by which dissolved species move rel-
1ffii1: Programmed computing: aids (Appendix C) could help solve problems marked atively slowly from areas of higher concentration to areas of lower concentration,
with this icon.